Icarus 214 (2011) 510–533
Contents lists available at ScienceDirect
Icarus
journal homepage: www.elsevier.com/locate/icarus
Saturn’s tropospheric composition and clouds from Cassini/VIMS 4.6–5.1 lm
nightside spectroscopy
Leigh N. Fletcher a,⇑, Kevin H. Baines b, Thomas W. Momary c, Adam P. Showman d, Patrick G.J. Irwin a,
Glenn S. Orton c, Maarten Roos-Serote e, C. Merlet a
a
Atmospheric, Oceanic & Planetary Physics, Department of Physics, University of Oxford, Clarendon Laboratory, Parks Road, Oxford OX1 3PU, UK
SSEC, University of Wisconsin-Madison, 1225 W. Dayton Street, Madison, WI 53706, USA
Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA
d
Department of Planetary Sciences, Lunar and Planetary Laboratory, University of Arizona, Tucson, AZ 85721, USA
e
Lisbon Astronomical Observatory, Tapada da Ajuda, 1349-018 Lisbon, Portugal
b
c
a r t i c l e
i n f o
Article history:
Received 28 February 2011
Revised 1 June 2011
Accepted 6 June 2011
Available online 8 July 2011
Keywords:
Saturn
Atmospheres, Composition
Atmospheres, Structure
a b s t r a c t
The latitudinal variation of Saturn’s tropospheric composition (NH3, PH3 and AsH3) and aerosol properties
(cloud altitudes and opacities) are derived from Cassini/VIMS 4.6–5.1 lm thermal emission spectroscopy
on the planet’s nightside (April 22, 2006). The gaseous and aerosol distributions are used to trace atmospheric circulation and chemistry within and below Saturn’s cloud decks (in the 1- to 4-bar region).
Extensive testing of VIMS spectral models is used to assess and minimise the effects of degeneracies
between retrieved variables and sensitivity to the choice of aerosol properties. Best fits indicate cloud
opacity in two regimes: (a) a compact cloud deck centred in the 2.5–2.8 bar region, symmetric between
the northern and southern hemispheres, with small-scale opacity variations responsible for numerous
narrow light/dark axisymmetric lanes; and (b) a hemispherically asymmetric population of aerosols at
pressures less than 1.4 bar (whose exact altitude and vertical structure is not constrained by nightside
spectra) which is 1.5–2.0 more opaque in the summer hemisphere than in the north and shows an
equatorial maximum between ±10° (planetocentric).
Saturn’s NH3 spatial variability shows significant enhancement by vertical advection within ±5° of the
equator and in axisymmetric bands at 23–25°S and 42–47°N. The latter is consistent with extratropical
upwelling in a dark band on the poleward side of the prograde jet at 41°N (planetocentric). PH3 dominates the morphology of the VIMS spectrum, and high-altitude PH3 at p < 1.3 bar has an equatorial maximum and a mid-latitude asymmetry (elevated in the summer hemisphere), whereas deep PH3 is
latitudinally-uniform with off-equatorial maxima near ±10°. The spatial distribution of AsH3 shows similar off-equatorial maxima at ±7° with a global abundance of 2–3 ppb. VIMS appears to be sensitive to
both (i) an upper tropospheric circulation (sensed by NH3 and upper-tropospheric PH3 and hazes) and
(ii) a lower tropospheric circulation (sensed by deep PH3, AsH3 and the lower cloud deck).
Ó 2011 Elsevier Inc. All rights reserved.
1. Introduction
The Visual and Infrared Mapping Spectrometer (VIMS, Brown
et al., 2004) onboard the Cassini spacecraft exploits a unique region
of Saturn’s spectrum between 4.6 and 5.1 lm where the effects of
scattered sunlight diminish; the collision-induced opacity due to
H2–He is at a minimum and strong CH4 absorptions are absent.
As a result, this wavelength range allows Cassini to probe deeper
into Saturn’s troposphere than at any other infrared wavelength.
As on Jupiter, this 5-lm window is sensitive to the emission of
the gas giant’s internal heat, attenuated by overlying cloud decks
⇑ Corresponding author.
E-mail address: fl[email protected] (L.N. Fletcher).
0019-1035/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.icarus.2011.06.006
that appear in silhouette against the warm thermal emission. To
date, analysis of VIMS data has focussed on the detailed morphology of images at discrete near-IR wavelengths (e.g., Baines et al.,
2006, 2009; Choi et al., 2009), which has revealed a wealth of information about dynamical phenomena within Saturn’s cloud decks
(e.g., strings of pearls, ribbon waves, the hexagon, polar vortices,
annular clouds, and equatorial plumes; see the review by Del Genio
et al. (2009)). However, the wavelength dependence of Saturn’s
4.6–5.1 lm spectrum (1950–2220 cm1) has yet to be fully
exploited. In this paper, we study the influences of gaseous distributions and cloud properties on nightside VIMS spectra (i.e., sensitive to thermal emission alone, in the absence of reflected sunlight)
to determine the latitudinal distribution of opacity sources in Saturn’s troposphere.
511
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
Saturn’s 5-lm window is expected to be similar to Jupiter’s, albeit with a different vertical distribution of tropospheric aerosols
due to Saturn’s lower gravity. Voyager/IRIS and Galileo/NIMS
investigations demonstrated that Jupiter’s 5-lm emission was
anticorrelated with both the visible albedo and with a variable
opacity cloud in the 1–2 bar region (e.g., Westphal et al., 1974; Terrile et al., 1977; Marten et al., 1981; Bézard et al., 1983; Irwin et al.,
1998; Roos-Serote et al., 1998; Irwin and Dyudina, 2002). This correlation is not readily apparent on Saturn, where the visible belt/
zone contrasts are subdued by the upper tropospheric hazes. Nevertheless, VIMS 5-lm images show extremely detailed zonal organisation, with a diverse range of meteorological features (fine-scale
zonal lanes, small vortices and other turbulent structures), some of
which are common to both visible and 5-lm imaging (e.g., Choi
et al., 2009; Vasavada et al., 2006).
Orton et al. (2009) reviewed the first investigations of Saturn’s
5-lm window from ground-based and space-based platforms,
starting with the first spectroscopic detections of CH3D (Fink and
Larson, 1978) and phosphine (PH3) by Larson (1980) from the Kuiper Airborne Observatory. PH3 was found to dominate the shape of
Saturn’s 5-lm emission, but its poorly understood absorption coefficients hampered quantitative analyses of the 5-lm window for
many years (Noll and Larson, 1990). The abundance of PH3 has
since been studied at 5 lm using a range of techniques (Bézard
et al., 1989; Noll and Larson, 1990; de Graauw et al., 1997).
Ground-based observations began to reveal the other principal
contributors to the 5-lm spectrum: CO was first detected in UK
Infrared Telescope measurements (UKIRT) at 5 lm by Noll et al.
(1986); germane (GeH3) from UKIRT (Noll et al., 1988) and later
ISO (de Graauw et al., 1997); and arsine (AsH3) from UKIRT (Noll
et al., 1989) and the Canada–France–Hawaii Telescope (Bézard
et al., 1989). NH3 bands (2m2 and m4) affect the long-wavelength
edge of this window and were first detected by Fink et al. (1983),
and later refined by Voyager/IRIS (Courtin et al., 1984) and ISO
(de Graauw et al., 1997). Detection of a subsolar H2O distribution
at 5 lm had to wait for disc-averaged ISO spectra in the 1990s
(de Graauw et al., 1997). Finally, IRTF imaging at 5.1 lm indicated
that Saturn’s deep cloud layers were spatially inhomogeneous
(Yanamandra-Fisher et al., 2001) before Cassini’s arrival. Although
the spectral resolution of VIMS is necessarily smaller than groundbased instruments, it offers the capability to map the spatial distribution of some of these gases for the first time, without having to
correct for telluric contamination.
Besides the wide ranging spectral effects of PH3, Saturn’s
poorly-understood cloud properties further complicate quantitative analyses of the VIMS spectra. The expected condensation altitudes for volatiles can be estimated using thermochemical
equilibrium theory and knowledge of bulk elemental abundances
(Weidenschilling and Lewis, 1973; Atreya et al., 1999), although
these do not account for mixing via atmospheric motions. Assuming a fivefold enhancement in concentrations over solar composition, calculations by Atreya et al. (1999) suggested that VIMS
observations probe vertical dynamics and chemistry in the NH3
(base at 2 bar) and NH4SH (base at 6 bar) ice cloud-forming regions
of Saturn’s troposphere. Our present knowledge of Saturn’s clouds,
largely derived from visible and near-IR reflectivity studies, is reviewed by West et al. (2009). Common features of the numerous
studies (e.g., Karkoschka and Tomasko, 1992; Karkoschka and
Tomasko, 1993; Stam et al., 2001; Temma et al., 2005; Pérez-Hoyos
et al., 2005; Karkoschka and Tomasko, 2005) include (a) a stratospheric haze (1 < p < 90 mbar) of small radius (r 0.1–0.2 lm) particles, presumably originating from photochemical processes; (b) a
tropospheric haze from the tropopause down to the first condensation cloud deck at 1.5–2.0 bar, possibly with aerosol-free gaps in
the vertical distribution; and (c) a possible thick NH3 cloud,
although no spectroscopic signature for NH3 ice has been observed.
As we shall demonstrate in Section 4, VIMS is sensitive to a combination of these upper level ubiquitous hazes and the deeper cloud
decks.
The spatial distribution of NH3 gas is intimately tied to the latitudinal variability of the hazes. Global constraints on the NH3 vertical distribution have been provided by a number of authors, as
highlighted in Table 1. Generally, NH3 was found to be around
500 ppm below 3 bar (de Pater and Massie, 1985; Briggs and Sackett, 1989), decreasing to 100 ppm at the condensation altitude
(Briggs and Sackett, 1989; Grossman et al., 1989; de Graauw
et al., 1997; Orton et al., 2000; Burgdorf et al., 2004) and then
decreasing with altitude according to a sub-saturated vapour pressure profile and photolysis in the upper troposphere (e.g., de Graauw et al., 1997; Kerola et al., 1997; Kim et al., 2006; Fletcher
et al., 2009b). In this work we derive the latitudinal distribution
of gaseous composition (NH3, PH3, AsH3) and cloud opacity (tropospheric clouds and hazes) from VIMS observations of the 5-lm
window. Section 2 describes the selection and error sources in
the VIMS data; Section 3 introduces the spectral model, techniques
and opacity sources allowing us to retrieve atmospheric properties.
The degeneracies between assumed cloud distributions and properties is explored in Section 4. Section 5 presents the VIMS-derived
distributions of gases and clouds and Section 6 describes their
implications for Saturn’s tropospheric dynamics and chemistry.
2. Observations
2.1. VIMS data and calibration
Saturn’s emitted radiance in the 4.6–5.1 lm region is measured
by the Visible and Infrared Mapping Spectrometer (VIMS, Brown
Table 1
Vertical distribution of ammonia mole fraction from previous determinations.
Reference
Courtin et al. (1984)
de Pater and Massie (1985)
Briggs and Sackett (1989)
Grossman et al. (1989)
Noll and Larson (1990)
de Graauw et al. (1997)
Kerola et al. (1997)
Orton et al. (2000)
Burgdorf et al. (2004)
Kim et al. (2006)
Fletcher et al. (2009a)
qNH3
Method
4
(0.5 2.0) 10
5 104 at p > 3 bar
3 105 at p < 1.25 bar
0.7 1.1 104 at p = 2 bar
1.2 104 around condensation level
Upper limit 3 104
1.1 104 at p = 1.2 bar
Less than 1 109 at radiative-convective boundary
1 104 with 3–4 uncertainty
1 104
6 108 at 460 mbar
3 108 at 390 mbar
(3.3 ± 0.3) 107 at 690 mbar
Voyager/IRIS 180–300 cm1
Very Large Array (VLA)
Radio TB
VLA
5 lm spectra
ISO/SWS
3 lm data
Sub-mm PH3 analysis
ISO/LWS 96–101 cm1
3 lm data
Cassini/CIRS far-IR
512
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
et al., 2004) on the Cassini spacecraft. Although this instrument
actually consists of two bore-sighted grating spectrometers, only
the infrared channel (0.85–5.1 lm) is considered in the present
study. VIMS has a passively cooled linear array of 256 InSb photodiode detectors operating at 55–60 K. VIMS-IR records spectral
images by stepping a 2-axis scan mirror orthogonally in the
along-slit (64 pixel positions of the visible channel slit) and
cross-dispersion directions. One spectrum is acquired at each of
64 mirror steps in the cross-dispersion direction, yielding an effective pixel size of 0.5 mrad on a 64 64 pixel grid. The near-IR spectral resolution is approximately 15 nm, sampled at intervals of
16.6 nm.
IR image cubes (two spatial and one spectral dimension) were
geometrically and photometrically calibrated (including despiking
and flat-fielding with 2005 calibration files) by the VIMS Science
team at the University of Arizona. The VIMS calibration procedure
was previously described by McCord et al. (2004), although uncertainties in absolute calibration have not been fully documented
(Sromovsky et al., 2010b). Systematic errors from pre-flight calibration are thought to be as large as 10% in regions of strong telluric H2O absorption, although random noise is expected to be
considerably smaller (less than one digital quantisation number,
corresponding to approximately 0.1% of the typical 5-lm radiance). Radiometrically-calibrated VIMS-IR Images were navigated
by reconstructing ‘backplanes’ from the post-observation Cassini
Mission SPICE kernels generated by NASA/JPL (i.e., information on
latitude and longitude, as well as incidence, emission, azimuthal
and phase angles) using the ISIS3 (Integrated Software for Imaging
Spectrometers) package provided by USGS (Gaddis et al., 1997).
Artefacts in VIMS spectra identified by Sromovsky et al.
(2010b), particularly those associated with responsivity corrections near overlaps between order sorting filters, are believed to
have no effect on the 4.6–5.1 lm spectrum (e.g., Fig. 8 of Brown
et al., 2004). Light potentially scattered within the spectrometer
has also been identified as a source of enhanced reflectivity in
low-signal regions (Sromovsky et al., 2010b), although no such
discrepancies between data and models have been identified in
the 5-lm window. Finally, we found no evidence for a shift in
wavelengths from the nominal grid for any of the image cubes used
in this study. Nightside VIMS-IR radiances were assigned uncertainties by considering the larger of (i) 12% of the radiance
measured by each pixel, or (ii) 12% of the mean radiance in the
4.6–5.1 lm range. This avoided unequal weightings of retrievals
to the low-signal regions of the Saturn spectrum (see Section 3).
The 12% envelope is conservative, adding quadrature-estimated errors due to pre-flight calibration as well as forward-model uncertainties on spectral line data. Specifically, we assumed that
systematic errors dominate the error budget.
2.2. Data selection
Reflected sunlight observations of the giant planets are complicated by the uncertain optical properties (shape, size distribution,
composition, phase function, opacity) of their cloud and haze
layers. To minimise these effects, we considered only VIMS nightside spectra at a sufficient distance from the day/night terminator
to ignore scattered sunlight. Scattered sunlight from Saturn’s rings
is unlikely to contaminate the nightside Saturn spectra, as water
ice in the rings has a low albedo at all VIMS wavelengths beyond
2.8 lm, and is particularly dark near 3 and 5 lm (Cuzzi et al.,
2009). Thermal emission from the atmosphere, in addition to
absorption and scattering processes, should determine the overall
shape of the 4.6–5.1 lm spectrum. This study uses eight VIMS-IR
image cubes from sequence VIMS_023SA_MIRMAPB010 (part of
sequence S20) on April 22, 2006 (Table 2). Saturn subtended 3.1°
during these observations, at a distance of 2.2 million km (38 Saturn radii). The relatively large spacecraft range meant that almost
the entirety of Saturn was captured within the 32 32 mrad field
of view, allowing multiple latitudes to be covered in a single cube
(from 40°S to 70°N). Saturn’s sub-solar latitude was 17.6°S during
these observations (a heliocentric longitude, Ls = 317.3°) approaching the southern autumnal equinox. As such, seasonal hemispheric
asymmetries in cloud colouration and atmospheric temperatures
(Fletcher et al., 2010) were still present.
The eight VIMS cubes sampled Saturn during an entire 10-hour
rotation (Table 2) so that a composite image from these cubes covered 360° of longitude (Fig. 1). The longitudinal displacement of
individual features over the 10-hour sequence was not accounted
for in the reprojections, and this is particularly apparent in the
overlap region of the first and last cubes in Table 2 (120°W). Four
wavelengths (4.6–5.1 lm) are displayed to demonstrate that atmospheric features appear similar across the spectral range, and that
an asymmetry between the northern and southern mid-latitudes
persisted in April 2006 (Baines et al., 2006). Indeed, the map at
4.6 lm (Fig. 1d) shows a well-defined boundary at 10°N between
the bright north and dark south, and that the equatorial zone is largely indistinguishable from the rest of the southern hemisphere at
this wavelength.
The dark equatorial zone is bordered by two regions of diffuse
emission between ±5°. This axisymmetric band is colocated with
the narrow prograde jet identified by Garcı´a-Melendo et al.
(2010), which exists in addition to the broad equatorial jet. An
irregular chain of dark features (referred to as equatorial plumes)
impinge on these diffuse regions from both north and south.
Mid-latitudes between ±5 and ±32° show the strongest asymmetry
between the hemispheres, with the northern hemisphere considerably brighter than the south. Both hemispheres are characterised
by a series of latitudinally-narrow bright and dark lanes, similar
to those observed in reflected sunlight (Vasavada et al., 2006; Choi
et al., 2009). Some discrete features are observed at off-equatorial
latitudes (particularly in the bands between 20° and 30° in both
hemispheres), although the spatial resolution of the S20 sequence
of images is insufficient to characterise small-scale features such as
the String of Pearls at 33°N (?; Choi et al., 2009).
Zonal mean radiances were extracted from the reprojected
maps onto two different meridional grids: (i) a coarse grid with a
step size and latitude width of 5° for preliminary testing; and (ii)
a fine grid with a size and width of 1° for the final zonal profiles.
Table 2
VIMS-IR cubes used in this study. The quoted longitude is for System III West at the start time of the observations.
Cube
Date
Start time (UTC)
Stop time (UTC)
Longitude
Range (km)
Phase (deg)
CM_1524383985
CM_1524388848
CM_1524393612
CM_1524400806
CM_1524403247
CM_1524408018
CM_1524412815
CM_1524417617
2006-April-22
2006-April-22
2006-April-22
2006-April-22
2006-April-22
2006-April-22
2006-April-22
2006-April-22
07:29:22
08:50:25
10:09:49
12:09:43
12:50:24
14:09:55
15:29:52
16:49:54
08:04:01
09:25:04
10:44:28
12:44:22
13:25:03
14:44:34
16:04:31
17:24:33
128.1
173.5
218.1
285.3
308.1
352.8
37.6
82.4
2,806,258
2,793,264
2,780,503
2,760,982
2,754,393
2,741,313
2,728,032
2,714,716
114
114
114
113
113
113
113
113
513
Planetocentric Latitude
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
μW/cm2/sr/μm
60
50
40
1.0
30
0.8
20
0.6
10
0
0.4
-10
0.2
-20
-30
360
0.0
340
320
300
280
260
240
220
200
180
160
140
120
100
80
60
40
20
0
Planetocentric Latitude
System III West Longitude
μW/cm2/sr/μm
60
50
40
0.6
30
0.5
20
0.4
10
0.3
0
-10
0.2
-20
0.1
-30
360
0.0
340
320
300
280
260
240
220
200
180
160
140
120
100
80
60
40
20
0
Planetocentric Latitude
System III West Longitude
μW/cm2/sr/μm
60
50
0.5
40
30
0.4
20
0.3
10
0
0.2
-10
0.1
-20
-30
360
0.0
340
320
300
280
260
240
220
200
180
160
140
120
100
80
60
40
20
0
Planetocentric Latitude
System III West Longitude
μW/cm2/sr/μm
60
50
40
0.4
30
20
0.3
10
0.2
0
-10
0.1
-20
-30
360
0.0
340
320
300
280
260
240
220
200
180
160
140
120
100
80
60
40
20
0
System III West Longitude
Fig. 1. Four examples of the VIMS-IR radiances at (a) 5.1; (b) 5.0; (c) 4.8 and (d) 4.6 lm. Radiances from eight cubes in Table 2 were reprojected onto a cylindrical map,
selecting only regions that were well separated from the day/night terminator. No attempt has been made to correct for the motion of cloud features due to the zonal flow
(see main text), resulting in some apparent disconnect at the overlap points. We have not corrected for limb darkening in these images, resulting in the visible seams where
the cubes overlap. Thermal emission was obscured by the noise at wavelengths shorter than 4.5 lm.
Within each latitude bin, we selected spectra within 10° of the
minimum emission angle for the latitude, and restricted selections
to phase angles greater than 90° and solar angles greater than 120°
– i.e., ensuring that only nightside observations contributed to the
average. The hemispheric asymmetry can be clearly seen in the
spectral comparison in Fig. 2. Radiances and brightness temperatures for the 4.6–5.1 lm region are compared for five latitudes,
showing that the northern mid-latitudes were uniformly 10–12 K
brighter than southern mid-latitudes in April 2006. The overall
morphology of the spectrum is dominated by absorption from
PH3 gas and tropospheric aerosols, although the band centres for
a variety of gases are labelled in Fig. 2b. Note that the broad
absorption feature at 4.74 lm is a blend of absorptions due to
PH3, GeH4, AsH3 and CO. The unusual ‘kink’ in the equatorial spectrum at 5.1 lm that is absent from other latitudes is a signature of
tropospheric NH3.
514
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
(a) Zonally-Averaged Nightside Radiance
1.0
30N
15N
EZ
15S
30S
Radiance (μW/cm2/sr/μm)
0.8
0.6
0.4
0.2
0.0
4.5
4.6
4.7
4.8
4.9
5.0
5.1
Wavelength (μm)
(b) Brightness Temperature
190
30N
15N
EZ
15S
30S
Brightness T (K)
180
170
PH3 ν2+ν4
PH3 ν1,ν3
NH3
160
CH3D ν2
PH3 2ν2
150
CO
AsH3
GeH4 ν3,ν1
140
4.5
4.6
4.7
4.8
4.9
5.0
5.1
Wavelength (μm)
Fig. 2. Comparisons of zonally-averaged radiances (a) and brightness temperatures (b) for five latitudes (the equator, ±15° and ±30°) extracted from the image cubes in Fig. 1.
Prominent features in the spectrum are labelled in (b) in their approximate locations, but these gases actually have effects over a wider region of the low-resolution VIMS
spectrum than indicated here. Radiance errors described in the main text are indicated in (a).
3. Spectral modelling
Fig. 2 showed that VIMS-IR spectra in the 4.6–5.1 lm range are
sensitive to a wide variety of gases and aerosols, but that the spectral resolution (approximately 15 nm, or R = k/Dk 330 at 5 lm) is
insufficient to resolve the individual lines. Instead, they blend together into absorption complexes, requiring simultaneous modelling of the entire range to derive the best-fitting atmospheric
profile at each latitude. In this section we describe the basic spectral model before exploring the degeneracies associated with the
VIMS spectra in Section 4.
3.1. Reference atmosphere
Saturn’s a priori atmospheric structure (temperatures, T(p), and
mole fractions, q(p)) was defined on a grid of 39 levels equally
spaced in logp between 10 mbar and 10 bar. Temperatures at each
latitude were obtained from Cassini/CIRS T(p) profiles from Cassini’s prime mission (sensitive to the 1–800 mbar range, Fletcher
et al., 2010), and extrapolated between 0.8 and 10 bar with a dry
adiabatic lapse rate, g/cp (where g is the latitudinally-variable gravitational acceleration at 1 bar and cp is the specific heat capacity of
Saturn’s H2–He–CH4 atmosphere).
Collision-induced absorption of H2–H2, H2–He, H2–CH4 and
CH4–CH4 was pre-calculated from the tabulations of Borysow
(1991, 1993), Borysow et al. (1988), Borysow and Frommhold
(1986, 1987) and references therein. The helium mixing ratio He/
H2 was set to 0.135 (Conrath and Gautier, 2000). Methane and its
isotopologues are well-mixed throughout the altitude range of
interest, and were included with mole fractions of 4.7 103
(CH4); 3.0 107 (CH3D) and 5.1 105 (13CH4) following Fletcher
et al. (2009b). The PH3 profile was set to the CIRS-derived abundance of 6.4 ppm at p > 0.55 bar, decreasing due to photolysis at
lower pressures with a fractional scale height of 0.27 (the ratio of
the PH3 scale height to the scale height of the bulk atmosphere,
Fletcher et al., 2009a). The vertical distribution of NH3 had a deep
mole fraction of 60 ppm (Fletcher et al., 2009b), decreasing with
altitude following a saturated vapour pressure profile (p > 0.3 bar)
and a linear extrapolation to low pressures to represent photolysis
(p < 0.3 bar). Minor constituents affecting the 4.6–5.1 lm range
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
were assumed to be well-mixed with altitude, and were included
with the following mole fractions: CO (1 ppb, Noll and Larson,
1990); GeH4 (0.4 ppb, Noll and Larson, 1990); AsH3 (3.0 ppb, Noll
and Larson, 1990); and H2O (well-mixed at 0.176 ppm over the altitude range of interest, de Graauw et al., 1997).
3.2. Sources of line data
The near-infrared spectral line database was updated from that
used by Irwin et al. (1998) and Roos-Serote et al. (1998) for Galileo/
NIMS analysis, who predominantly used line data extracted from
GEISA 1992 (Husson et al., 1992). HITRAN2004 (Rothman et al.,
2005) was used for CO, H2O, CH4, CH3D and 13CH4. Absorption
due to phosphine’s pentad polyad dominates the VIMS 5-lm spectrum, with the 2m2 band at 5.07 lm and the broad m2 + m4 band between 4.69 and 4.78 lm. Furthermore, the m1 and m3 bands absorb
shortward of 4.58 lm and contribute to the reduced thermal emission at these wavelengths. GEISA2003 (Jacquinet-Husson et al.,
2005) was used for PH3 as it contained updates from Kleiner
et al. (2003) for some missing bands in the 5-lm window. However, the PH3 absorption coefficients are still subject to considerable uncertainty, as the original intensity studies of Tarrago et al.
(1992) are estimated to have only a 20–30% accuracy. Work is
underway to compare this band to the dyad at 9 lm (Fusina and
Di Lonardo, 2000; Brown et al., 2002) and the octad at 2.9 lm (Butler et al., 2006). GEISA2003 was also used for GeH4, and contained
updated 0–5300 cm1 line data for NH3 from Kleiner et al. (2003).
AsH3 was not present in either database, so we used line data from
Dana et al. (1993) and Mandin and Aug. (1995), following the NIMS
analyses of Irwin et al. (1998) and Roos-Serote et al. (1998).
Foreign broadening (by H2) for each of the molecules was
estimated for all lines as follows. GeH4 was broadened with a
half-width of 0.1 cm1 atm1 and a temperature dependence T0.75
(Jacquinet-Husson et al., 2005). AsH3 had a half width of
0.075 cm1 atm1 and T0.5 (an assumption based on PH3). PH3 used
estimated half-widths from Kleiner et al. (2003) and T0.65. NH3 had
a half-width of 0.072 cm1 atm1 and T0.73 (B. Bézard and L. Brown,
personal communication). The spectroscopic data for each gas
were used to generate k-distributions (ranking absorption coefficients, k, according to their frequency distribution, Irwin et al.,
2008) using a 16 nm FWHM on an evenly sampled wavelength grid
of 8 nm spacing. We use a direct sorting method to calculate the kdistribution from line-by-line spectra within each spectral bin (e.g.,
Goody et al., 1989). A triangular instrument function was used for
spectral modelling, which was found to be a good approximation
for grating spectrometers with rectangular entrance slits and linear
arrays of detectors, and allows rapid convolution over the k-distribution. The use of pre-tabulated k-distributions greatly accelerates
spectral calculations and permits rapid retrieval of atmospheric
spectra.
515
estimation (Rodgers, 2000), but adapted for planetary applications
by tuning a priori uncertainties to achieve the optimal trade-off between precision (the quality of the spectral fit to the data) and
physically-realistic solutions (Irwin et al., 2008). Retrievals require
calculations of both the upwelling radiance, I(k), as well as the rate
of change of radiance with the model parameters (dI/dx), based on
the reference atmosphere (Section 3.1), which is perturbed in successive iterations (based on a Marquardt–Levenburg braking
parameter, Press et al., 1992) to converge on the optimal solution.
The algorithm seeks to minimise the residual between measured
and modelled spectra (the traditional v2).
The 5-lm window can be modelled assuming thermal emission
from the planet, attenuated by absorbing clouds. We neglect any
thermal emission from the clouds themselves, as these reside at
higher, cooler (by 50–80 K) altitudes than the source of the upwelling radiance (the 4–6 bar region, where temperatures reach
approximately 240 K, see Section 3.5). In this case the functional
derivatives (or Jacobians, dI/dx) are computed analytically, permitting rapid convergence to the optimal solution. However, multiple
scattering from aerosols in the real saturnian atmosphere will increase the optical paths of individual photons, thereby enhancing
the gas absorptions and decreasing the molecular abundances required to reproduce the spectra.
The Nemesis software performs full multiple-scattering calculations (either for thermal emission, reflected sunlight or a combination of both) using a matrix operator (or doubling–adding)
approach (Plass et al., 1973; Hansen and Travis, 1974) in a planeparallel atmosphere, but numerical-differencing must be used to
evaluate the functional derivatives due to the complexity of the
multiple-scattering scheme. Integration of the scattered radiance
over all solid angles was simplified in two ways: first, the integration over zenith angle used a Lobatto quadrature scheme with five
angles to reduce the calculation to a simple weighted sum. The
scattering scheme must use sufficient zenith angle quadrature
points to represent the phase functions of the scattering particles.
Second, as thermal scattering is an azimuthally symmetric process,
only the first (azimuthally-independent) Fourier component was
used. The numerical calculations involved in multiple scattering
are computationally expensive, slowing the retrieval process by
an order of magnitude, but has a significant effect on the 5-lm
window of the VIMS spectrum (Section 4).
3.4. Introducing cloud models
As knowledge of Saturn’s vertical cloud structure and optical
properties remains rather limited, we aimed to explore a broad
range of parameter space with a variety of different cloud models
(Table 3). A full vertical opacity retrieval was poorly constrained by
the 5-lm data due to the degeneracy between PH3 and aerosols.
Instead, four parameterised vertical structure models were considered (optical depths are quoted for 5 lm);
3.3. Forward modelling and retrieval
VIMS spectra were analysed using a suite of radiative transfer
and retrieval codes developed at the University of Oxford (Nemesis, Irwin et al., 1997, 2008), which have been previously used to
investigate Galileo/NIMS near-IR spectra of Jupiter (e.g., Irwin
et al., 1998; Irwin and Dyudina, 2002) and Cassini/CIRS thermalIR spectra of Jupiter and Saturn (e.g., Fletcher et al., 2009a, 2010).
The correlated-k method (Goody et al., 1989; Lacis and Oinas,
1991) is used for rapid calculation of non-monochromatic transmission along an inhomogeneous atmospheric path based upon
pre-tabulated absorption coefficients, aerosol extinction cross-sections and collision-induced absorption. Retrievals of temperature,
aerosol and gaseous composition are achieved using optimal
I: Single compact cloud: A single aerosol layer with variable optical thickness s1 and base pressure, pb.
II: Two compact clouds: A compact aerosol layer with a variable
s1, composition and base pressure was placed beneath a
spectrally-grey cloud at a fixed altitude with variable opacity,
s2. This upper cloud was arbitrarily placed at the predicted
NH3 condensation altitude for a solar nitrogen abundance
(1.47 bar, 152 K, Atreya et al., 1999) to minimise the number
of free parameters in the model, although it would be deeper
for bulk enrichments in Saturn’s nitrogen content.
III: Single extended cloud: A well-mixed distribution of aerosols
with variable opacity s1 between the 100-mbar pressure level
(the tropopause) and a variable base pressure, pb.
516
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
Phase Function at 5.0 μm
Table 3
Summary of cloud models tested in this study, vertical structures I–IV, optical models
A–E.
Cloud
model
Description
Variables and comments
I
II
Single compact cloud
Two compact clouds
III
Single extended cloud
IV
A
Compact upper,
extended deep
Grey cloud
B
C
D
NH3 ice
NH4SH
Modified pseudo-NH4SH
E
Updated NH4SH
pb, s1
pb, s1, s2
(Grey upper cloud fixed at 1.5 bar)
pb, s1
(Extends to the tropopause)
pb, s1, s2
(Grey upper cloud fixed at 1.5 bar)
Grey cross-section and x0 = 0.95
Across the full range; isotropic phase
function
Martonchik et al. (1984)
Ferraro et al. (1980)
Refractive index 2.3 + 0.01i
Nixon et al. (2001)
Howett et al. (2007)
IV: Compact upper cloud, extended deep cloud: A combination of
the physically-thin upper cloud from Model II and the extended
cloud from Model III.
In addition to the vertical structure, we tested the sensitivity to
the aerosol composition by calculating extinction cross-sections
and phase functions p(h) based on the refractive indices in Table 3
and shown graphically in Fig. 3. The models tested were (A) a grey
cross-section and single scattering albedo (x0 = 0.95) across the
4.6–5.1 lm range, with an isotropic phase function (Fig. 4);
P(θ) at 5.0 μm
1.5
1.0
0.5
0
0.001
4.6
4.7
4.8
4.9
5.0
100
150
Fig. 4. Variation of phase function with scattering angle for the range of cloud
compositions used in this study. Phase functions were calculated as a two-term
Henyey–Greenstein (HG) functions based on the optical properties listed in the key.
With the exception of the isotropic scatterer, there is little to distinguish between
these phase functions, which is mostly governed by the choice of particle sizes
(r = 1.0 ± 0.05 lm).
(B) pure NH3 ice (Martonchik et al., 1984); (C) pure NH4SH (Ferraro
et al., 1980); (D) a modified pseudo-NH4SH cloud based on a refractive index of 2.3 + 0.01i (following suggestions by, Nixon et al.,
2001); and (E) updated NH4SH optical constants from Howett
et al. (2007). For each cloud type in Table 3, Mie theory was used
(b) Real Refractive Index
Real Refractive Index (n)
Imaginary Refractive Index (k)
0.010
50
Scattering Angle (θ)
(a) Imaginary Refractive Index
4.5
NH3 (Martonchik et al., 1984)
NH4 SH (Ferraro et al., 1980)
Grey Isotropic Cloud
NH4 SH (Howett et al., 2007)
N2 H4 (Clapp and Miller, 1996)
Pseudo-NH4 SH (Nixon et al., 2001)
2.0
5.1
2.4
2.2
NH3 (Martonchik et al., 1984)
NH4 SH (Ferraro et al., 1980)
Grey Isotropic Cloud
NH4 SH (Howett et al., 2007)
N2 H4 (Clapp and Miller, 1996)
Pseudo-NH4 SH (Nixon et al., 2001)
2.0
1.8
1.6
1.4
4.5
4.6
4.7
4.8
4.9
5.0
5.1
Wavelength (μm)
Wavelength (μm)
(c) Extinction Cross Section
(d) Single Scattering Albedo
Single Scattering Albedo
Cross section (10-7 cm2 )
1.2
1.0
0.8
0.6
0.4
0.2
0.0
4.5
4.6
4.7
4.8
4.9
5.0
Wavelength (μm)
5.1
1.00
0.95
0.90
4.5
4.6
4.7
4.8
4.9
5.0
5.1
Wavelength (μm)
Fig. 3. Optical constants for the range of cloud compositions considered in this study (a key for the different lines is shown in (b). Imaginary (a) and real (b) refractive indices
are taken from the listed references. Extinction cross-sections (c) and single scattering albedos (d) were calculated using Mie theory for a standard gamma distribution of
particles of radius r = 1.0 ± 0.05 lm. The key difference between the compositions is the enhanced absorption of NH4SH-like species (Ferraro et al., 1980; Howett et al., 2007)
between 4.7 and 4.9 lm.
517
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
to calculate the scattering properties of spherical 1-lm-radius particles with a standard gamma distribution of particle sizes, variance 0.05 lm. The phase function p(h) (shown in Fig. 4) was
calculated as a two-term Henyey–Greenstein (HG) function
(explicitly calculating the fraction of forward scattering and the
asymmetries in the forward and backward scattering functions).
With the exception of the isotropic scatterer, there is little to distinguish between the phase functions in Fig. 4, which are mostly
determined by the chosen particle size. The crucial differences between the cloud models lies in the wavelength-dependence of the
single scattering albedo (Fig. 3d, related to the imaginary refractive
indices in Fig. 3a).
The absorption cross-sections, HG phase functions and size distributions in Figs. 3 and 4 were not intended to be an exhaustive
representation of Saturn’s clouds, given the substantial degeneracies inherent in the interpretation of VIMS spectra. Nevertheless,
they are broadly representative of the types of condensed phases
that might be present in Saturn’s troposphere. Addition of further
(a) PH3
(b) NH3
0.6
0.4
0.2
4.7
4.8
4.9
5.0
0.8
0.6
0.4
0.2
0.0
4.6
5.1
1.0
Radiance (μW/cm2 /sr/μm)
0.8
4.7
Wavelength (μm)
(d) AsH3
0.6
0.4
0.2
4.7
4.8
4.9
5.0
0.6
0.4
0.2
4.7
4.8
4.9
5.0
0.2
5.0
Wavelength (μm)
5.1
0.2
4.7
0.6
0.4
0.2
4.8
4.9
4.9
5.0
5.1
(i) 13CH4
1.0
4.7
4.8
Wavelength (μm)
0.8
0.0
4.6
5.1
0.4
0.0
4.6
5.1
Radiance (μW/cm2 /sr/μm)
Radiance (μW/cm2 /sr/μm)
0.4
5.0
0.6
(h) CH3D
0.6
4.9
0.8
Wavelength (μm)
0.8
4.8
(f) CO
1.0
4.9
4.7
1.0
(g) CH4
4.8
0.2
Wavelength (μm)
0.8
0.0
4.6
5.1
1.0
4.7
0.4
0.0
4.6
5.1
Radiance (μW/cm2 /sr/μm)
Radiance (μW/cm2 /sr/μm)
Radiance (μW/cm2 /sr/μm)
5.0
0.6
(e) H2O
Wavelength (μm)
Radiance (μW/cm2 /sr/μm)
4.9
1.0
0.8
0.0
4.6
4.8
0.8
Wavelength (μm)
1.0
0.0
4.6
(c) GeH4
1.0
Radiance (μW/cm2 /sr/μm)
Radiance (μW/cm2 /sr/μm)
1.0
0.0
4.6
complexity (e.g., using the dual-absorber of NH3 and NH4SH following (Sromovsky et al., 2010a), or introduction of non-spherical
particles) was not warranted by the 4.6–5.1 lm data, but such a
combination is certainly possible for Saturn’s cloud decks. Hydrazine (N2H4), from the photolysis of tropospheric NH3, is not expected to be a major constituent of the tropospheric clouds, and
the single scattering albedo and phase function in Figs. 3 and 4
(Clapp et al., 1996) are not sufficiently different in the 4.6–
5.1 lm range to distinguish hydrazine from NH3 ice in the VIMS
spectrum. However, one important species is absent from Table 3
that could potentially be a major contributor to IR opacity in this
spectral range – diphosphine (P2H4), which is expected to be present in significant quantities from PH3 photolysis. Very little is
known about the absorptive and scattering properties of P2H4,
and a determination of the optical properties of diphosphene is
an urgent priority for future VIMS studies, particularly in reflected
sunlight. The uncertain spectral properties of non-spherical particles, NH3 + NH4SH mixes, P2H4 and other potential contaminants
5.0
Wavelength (μm)
5.1
0.10
0.50
1.0
2.0
4.0
5.0
10.
0.8
0.6
0.4
0.2
0.0
4.6
4.7
4.8
4.9
5.0
5.1
Wavelength (μm)
Fig. 5. Sensitivity of VIMS nightside spectra to a selection of gases in the model atmosphere. Spectra were calculated using a compact grey-absorbing cloud in the absence of
scattering, with molecular abundances scaled from 0.1 to 10 times the a priori values (key is shown in (i)). PH3, NH3 and AsH3 have the largest effects over this spectral range.
518
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
The vertical sensitivity of the spectra is highly dependent on the
scattering properties and opacity of the cloud layers as well as the
abundances of the absorbing gases. However, an estimate of the
sensitivity is provided by the functional derivatives for the best-fitting II.A model (Fig. 6, using compact clouds at 1.4 and 2.7 bar and
s1 = s2 = 1). The functional derivatives have been normalised, so
that no account has been made for the magnitude of their spectral
effects from Fig. 5. VIMS spectra are generally sensitive to abundance profiles in the 1–6 bar region, with peak sensitivity for
PH3, NH3 and AsH3 in the 1–3 bar range. Some gases (notably
GeH4, CO and CH3D) show sensitivity to the 0.4–1.0 bar range,
although these generally have a smaller overall effect on the spectrum. In the absence of absorbing cloud layers, contribution functions (Fig. 7, the product of the transmission weighting function,
ds/dz, and the black body emission, B(z, T)) demonstrate that Saturn’s thermal emission originates from the 4–6 bar region, where
atmospheric temperatures reach approximately 240 K. In the absence of absorbing/scattering aerosols, the radiance in the 5-lm
in Saturn’s clouds could significantly alter the retrieved opacities
and cloud altitudes described in Section 5. Nevertheless, useful latitudinal contrasts in atmospheric parameters can still be derived.
3.5. Sensitivity analysis
Synthetic spectra for each gas contributing to the 5-lm window
are presented in Fig. 5. The mole fractions in the reference atmosphere were scaled by arbitrary amounts to show their spectral
influence (using a simple compact grey cloud, model II.A). PH3,
NH3 and AsH3 have the largest contributions to this range, with
smaller influences from H2O, GeH4, CO and CH3D. PH3 in particular
has a strong effect on the mean flux and spectral gradient between
4.9 and 5.0 lm. Variations of CH4 and 13CH4 have negligible effects
on the spectra. Some of the spectral signatures are similar (e.g.,
AsH3 and GeH4, not to mention those of aerosol absorption and
the broad effects of PH3), leading to degeneracies in the interpretation of VIMS spectra, which will be explored below.
(b) PH3
(a) Temperature
4.7
4.8
4.9
5.0
0.0
4.7
Wavelength (μm)
4.8
(d) GeH4
4.9
5.0
4.7
4.8
4.9
5.0
0.0
0.2
0.4
0.6
0.8
1.0
4.7
4.8
4.9
5.0
4.7
0.2
0.4
0.6
0.8
5.0
5.1
1.0
Pressure (bar)
Pressure (bar)
Wavelength (μm)
4.9
(i) CH3D
1.0
10.0
4.6
4.8
0.1
0.0
5.1
5.1
Wavelength (μm)
(h) CH4
5.0
5.0
1.0
10.0
4.6
5.1
0.1
4.9
4.9
(f) H2O
Wavelength (μm)
1.0
4.8
Wavelength (μm)
1.0
(g) CO
4.8
4.7
0.1
10.0
4.6
5.1
0.1
4.7
1.0
10.0
4.6
5.1
Pressure (bar)
Pressure (bar)
Pressure (bar)
1.0
(e) AsH3
Wavelength (μm)
Pressure (bar)
0.8
0.1
1.0
10.0
4.6
0.6
Wavelength (μm)
0.1
10.0
4.6
0.4
1.0
10.0
4.6
5.1
0.2
Pressure (bar)
1.0
10.0
4.6
(c) NH3
0.1
0.1
Pressure (bar)
Pressure (bar)
0.1
4.7
4.8
4.9
5.0
Wavelength (μm)
5.1
1.0
10.0
4.6
4.7
4.8
4.9
5.0
5.1
Wavelength (μm)
Fig. 6. Examples of the functional derivatives (Jacobians, the rate of change of radiance with a particular model parameter) for temperature and several gases contributing to
the 5-lm window. Spectra were calculated using a compact grey-absorbing cloud (model II.A), so results will vary depending on the properties of the absorbing gases.
Jacobians have been normalised to unity for each gas, and this does not represent their relative contribution to the spectrum (see Fig. 5, for example). A scale bar is shown for
the central three panels. VIMS spectra are mostly sensitive to compositional variations in the 1–3 bar region.
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
Pressure (bar)
0.1
1.0
10.0
4.7
4.8
4.9
5.0
5.1
Wavelength (μm)
0.0
0.2
0.4
0.6
0.8
1.0
Fig. 7. Contribution function (product of the transmission weighting function and
the Planck black body function) calculated for a cloud-free case for the VIMS
spectrum. The contribution function shows a maximum sensitivity to the 5-bar
level.
window would therefore be considerably larger than the 140–
190 K brightness temperatures in Fig. 2.
4. Model degeneracies and validation
Modelling a single VIMS spectrum is relatively straightforward
– a large number of gaseous and aerosol parameters can be tuned
to provide an excellent fit to the low resolution VIMS spectra
(R 330 at 5 lm). However, the results must also be physically
realistic when multiple retrievals are performed to study Saturn’s
zonal mean properties. Furthermore, the residuals between measured and synthetic spectra (the v2 parameter) should be as spatially uniform as possible to ensure that we have captured all of
the variability in the model. This section will explore the degeneracies inherent in modelling VIMS nightside spectra in the absence
of prior constraints on Saturn’s aerosol optical properties and distributions. Through extensive tests of the model with different
temperature, composition and cloud assumptions with the 22
coarse zonal-mean spectra described in Section 2.2, we demonstrate that VIMS data can provide robust conclusions about relative
spatial variability, even if absolute abundances and opacities are
poorly constrained.
4.1. Model assumptions
We began by testing a number of assumptions in our forward
models and retrievals. The simplest solution would be to fit the
spectrum by varying T(p) or PH3 alone, in the absence of attenuating/scattering aerosols. However, thermal variations needed to be
unrealistically large in the 1–5 bar region to reproduce the cool
brightness temperatures observed in Fig. 2, and it proved impossible to reproduce the 4.6–4.9 lm and 4.9–5.1 lm regions simultaneously by varying PH3 alone. Furthermore, fixing all the gases at
their a priori distributions and varying the opacity of the simplest
cloud model (I.A, a single compact grey-absorbing cloud in Table 3)
failed to reproduce the spectrum. The VIMS spectra can only be
reproduced by a simultaneous retrieval of gaseous abundances
and aerosol opacity.
But which gases to include in the retrieval? With the best-fitting
aerosol distribution in the grey-absorbing case (both scattering
519
and non-scattering), we sequentially added scaled retrievals of
each gas (i.e., fixing the vertical profile but varying the absolute
abundance) and assessed (a) the quantitative improvement to v2
and (b) the qualitative appearance of the meridional distribution
from the 22 spectra. Variations of PH3 were essential, whereas
the importance of NH3 only became apparent once we investigated
equatorial latitudes where the spectrum near 5.1 lm appears
markedly different from other regions (Fig. 2). The addition of
AsH3 moderately improved the fit in the region surrounding the
broad absorption at 4.74 lm (this was especially true at low latitudes). However, although the remaining gases in the model
(GeH4, CO, H2O and CH3D) had some minor effects on the spectra,
they did not deviate far from their a priori abundances and were
deemed insignificant (using an F-statistic test, Bevington and Robinson, 1992). Omitting these four gases from the retrieval had negligible effects on the retrieved NH3, PH3 and AsH3 abundances.
Adding complex cloud parameterisations: Simultaneously fitting
for the spatial variation of PH3, NH3 and AsH3, along with the variable opacity and depth of the single-cloud model I.A failed to provide adequate fits to the VIMS spectra. A similar conclusion was
reached for Galileo/NIMS 5-lm spectra of Jupiter (e.g., Irwin
et al., 2001), which required two separate aerosol populations, suggestive (but not uniquely) of a main jovian cloud deck of NH4SH
overlain by optically thin NH3 clouds. This prompted the development of the three additional cloud models (II–IV) in Section 3.4
which immediately improved the fits to the VIMS spectra. The 2cloud schemes produced the best fits as the two opacity sources
were allowed to vary independently of one another, increasing
the number of free parameters available for the retrieval. In addition, experiments varying both the deep cloud base pressure (pb)
and opacity (s1) showed that they have sufficiently different spectral effects to make them separable. A comparison of the v2 values
in Section 4.2 for the four different vertical models show that,
while some can be ruled out, others gave such similar spectral results that they could not be distinguished from each other. Furthermore, the VIMS spectra are insensitive to location and extent of the
upper cloud in models II and IV – shifting the base pressure between 1.4 and 1.8 bar for both compact and extended upper clouds
had no effect on the fitted spectra, only the cumulative opacity (s2)
has an influence.
Temperature variations in the deep troposphere: Independent retrievals of T(p) from the 5-lm window would be impossible given
the degeneracy with PH3 and aerosols, although spatial variations
are expected to be small. Nevertheless, three different assumptions were tested: (a) a mean CIRS-derived T(p) from Cassini’s
prime mission (Fletcher et al., 2009a) with the same lapse rate g/
cp for all latitudes; (b) the same mean T(p) but with a latitudedependent lapse rate (i.e., varying with g); and (c) the full
meridional CIRS T(p) with a latitude-variable lapse rate. The last
assumption provided the best fits to the VIMS spectra (there is
VIMS sensitivity to p < 800 mbar in Figs. 6 and 7), although in practise there was little to differentiate between the three cases. Retrieved meridional distributions of NH3 and AsH3 were very
similar for all three assumptions, but PH3 and aerosol optical
depths were affected. North–south asymmetries of PH3 and aerosols were present for all three cases, but the PH3 asymmetry was
smaller when the CIRS-derived tropospheric temperature asymmetry was accounted for. Uncertainties in retrieved absolute values arising from the differing temperature assumptions are 13%,
16% and 10% for PH3, NH3 and AsH3, respectively; 33% and 50%
for the deep and upper cloud opacities. Thus the retrieved gaseous
composition and aerosols have a degeneracy with the deep atmospheric temperatures, but the best-fitting assumption (c) was used
for the remainder of this study.
Vertical distribution of PH3: Early models of VIMS spectra (e.g.,
Baines et al., 2009) assumed PH3 to be well-mixed up to 0.55 bar
520
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
(Fletcher et al., 2009a). However, it proved difficult to simultaneously fit the 4.6–4.7 lm radiances and the 4.8–5.0 lm spectral
gradient. Fixing the PH3 abundance at some mean for all latitudes
generally worsened the quality of the spectral fits, particularly over
the southern hemisphere (PH3 was noted to be elevated in the
southern troposphere, Fletcher et al., 2009a). Finally, we parameterised the vertical PH3 distribution in terms of a deep mole fraction (q0) up to a pressure level (p0), followed by a decreasing
abundance with altitude according to a fractional scale height (f,
the ratio of the gas scale height to that of the bulk atmosphere)
(Fletcher et al., 2007a). Varying p0 simultaneously with aerosols
and other gaseous mole fractions for all 22 VIMS spectra (in both
the scattering and non-scattering cases), we found optimum fits
for p0 between 1.1 and 1.3 bar (examples of the v2 surfaces for
the equator, 30°N and 30°S are shown in Fig. 8). The cloud base
and PH3 p0 were found to be at lower pressures (higher altitudes)
at the equator than at mid-latitudes.
The introduction of the parameterised PH3 profile had a substantial effect on the v2 at all latitudes, producing closer fits to both
the 4.6–4.7 lm region and the spectral gradient between 4.8 and
5.0 lm. Similar tests for NH3 and AsH3 indicated that the wellmixed assumption was perfectly valid. However, the PH3 p0 determined by VIMS was considerably deeper than that determined by
CIRS and sub-mm data (0.55–0.65 bar, Orton et al., 2000; Fletcher
et al., 2009a). Furthermore, the retrieved fractional scale height
was rather small, permitting negligible PH3 abundances at
p < 1 bar, again inconsistent with CIRS. Indeed, when we compared
the range of PH3 abundances derived from VIMS (3.0–5.5 ppm for
the range of cloud models studied here, Section 5) to that derived
from CIRS (5.4–8.2 ppm, with a mean of 6.4 ppm, Fletcher et al.,
2009a), we found that VIMS and CIRS PH3 abundances differed
by factors of 1.5–1.8, even though the meridional variations of q0
and f were similar from both instruments. Although CIRS is sensitive to lower pressures (300–800 mbar) than VIMS (2–3 bar), we
expect PH3 to be well-mixed between these two levels. It is unlikely that near-IR line strengths could be too strong by a factor of
two, as this exceeds the uncertainties on line data for either the
near-IR or mid-IR vibrational bands (Section 3.2). Identification of
the source of this discrepancy will require (i) consistent measurements of PH3 line data across multiple bands; (ii) higher spectral
resolution observation of Saturn’s emission to separate PH3
absorption from continuum opacity sources; and (iii) simultaneous
near and mid-IR retrievals in the presence of tropospheric aerosols.
Nevertheless, relative PH3 variations can still be derived from VIMS
spectra.
At the start of this analysis, it was hoped that VIMS nightside
spectra would constrain a unique cloud model and, independently,
the spatial distribution of gases. However, the degeneracies between the different model parameters soon became overwhelming. Fig. 9 shows the meridional distribution of v2 for all four
vertical models (I–IV), scattering and non-scattering cases, for optical models A–C (Figs. 3 and 4). Testing of models D (pseudo-NH4SH
cloud of Nixon et al. (2001)) and E (updated NH4SH optical constants by Howett et al. (2007)) produced negligible differences to
(i) the quality of the spectral fits and (ii) the meridional distributions of gases and aerosols, so were omitted from the subsequent
analysis. All cloud models produce poor fits poleward of 55°N
due to a failure of our models to fit the higher emission angles
(sensitive to higher altitudes in Saturn’s atmosphere).
In general, the compact cloud models I and II produced the best
fits to the spectra. The 2-cloud scheme fitted better at the equator
(Fig. 9); at latitudes poleward of 10°S and the 35–65°N region of
the northern hemisphere. However, the 2-cloud scheme cannot
be distinguished from the single cloud scheme between 10 and
35°N, in a region where haze opacity is thought to be negligible
(see Section 5). Finally, although the extended deep cloud models
(III and IV) produced reasonable fits to the spectrum by eye, the
v2 (Fig. 9) was sufficiently different to distinguish between the
compact and extended cloud structures for the deep cloud.
In the non-scattering case, the residuals for optical models A–C
(the grey, NH3 and NH4SH cloud compositions) were indistinguishable from one another. Modelling scattering within the clouds improved the fits for the grey isotropic scatterer and NH4SH clouds,
but not for NH3 (although differences in the chosen particle sizes
could have an effect on this conclusion). Ruling out pure NH3 ice,
the two remaining optical models produced very similar fits to
the VIMS spectra: the grey assumption was better at northern
mid-latitudes (10–40°N) whereas solid NH4SH provided a better
fit at all other latitudes (Fig. 9). Although this is certainly not a unique solution, the NH4SH imaginary refractive indices in Fig. 3 from
both Ferraro et al. (1980) and Howett et al. (2007) suggest smaller
single scattering albedos between 4.7 and 4.9 lm compared to NH3
ice or the other possible cloud compositions, even though the scattering phase functions (Fig. 4) are all very similar to one another.
This difference provides marginally better fits for solid NH4SH than
other constituents, although we have not conducted an exhaustive
study of possible cloud candidates (e.g., the omission of diphosphine described in Section 3.4). In conclusion, the compact
Lat: EZ
Lat: 30N
3.5
3.5
3.0
3.0
3.0
2.5
2.0
1.5
1.0
0.5
1
2
3
Cloud Base Pressure (bar)
4
PH3 Knee Pressure (bar)
3.5
PH3 Knee Pressure (bar)
PH3 Knee Pressure (bar)
Lat: 30S
4.2. Degeneracies in spectral modelling
2.5
2.0
1.5
1.0
0.5
1
2
3
Cloud Base Pressure (bar)
4
2.5
2.0
1.5
1.0
0.5
1
2
3
4
Cloud Base Pressure (bar)
Fig. 8. Contours of v2 for VIMS retrievals varying the base pressure pb of the deep cloud layer and the transitional pressure p0 from well-mixed deep PH3 to PH3 in the
photochemical depletion region. Three representative latitudes are shown, indicating that the best fitting p0 is 1.3 bar, although this can be at lower pressures at the equator.
521
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
A: Grey Cloud Chi-Squared
Chisq of Fits
1.5
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
1.0
0.5
0.0
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
B: NH3 Cloud Chi-Squared
Chisq of Fits
1.5
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
1.0
0.5
0.0
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
C: NH4SH Cloud Chi-Squared
Chisq of Fits
1.5
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
1.0
0.5
0.0
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
Fig. 9. The meridional distribution of v2 for optical models A–C (grey, NH3 and NH4SH cloud scattering properties), which depend on the scattering processes (T = nonscattering, S = scattering) and vertical models (I–IV). These retrievals were performed for 22 coarsely gridded VIMS zonal mean spectra using the grey-cloud approximation.
Comparison of these values was used to rule out certain optical models and vertical distributions, although considerable degeneracies still exist.
scattering cloud schemes (models I and II, with II doing better at
known ‘hazy’ latitudes), in combination with the grey or NH4SH
optical properties (A and C) provided the best reproductions of
the VIMS spectra.
4.2.1. Aerosol profile uncertainties
Given that no single cloud model provided the best fits to the
data, we have to consider the range of possible solutions to this
underconstrained problem. If compact cloud layers are used (models I–II), the base pressures for the opacity (pb) must be placed between 1.9 and 2.7 bar. In the southern hemisphere, where the
opacity of the upper cloud is highest, pb is poorly constrained
and can be at any pressure greater than 2 bar (see the v2 figures
at the top of Fig. 12). Extended well-mixed clouds in the deep troposphere (models III–IV, previously shown to give poor reproductions of the VIMS data) require pb > 2.8 bar, and typically place the
bottom of the cloud between 3.3 and 4.0 bar.
The addition of scattering to the model causes the retrieved
optical depths of the deep cloud (s1) and upper cloud (s2) to increase by factors of 2–5 relative to the non-scattering case
(depending on the chosen optical model, Fig. 10). Furthermore,
scattering introduces an emission-angle dependence to the deep
optical depths if the cloud is an isotropic grey scatterer
(Fig. 10a), but not when it is comprised of solid NH4SH (Fig. 10b).
As the emission angle is varying from equator to pole, this suggests
that the isotropic phase function (Fig. 4) is a poor representation of
522
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
(a) Deep Cloud Opacity - Grey Cloud
Optical Depth of Deep Cloud
8
6
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
4
2
0
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
(b) Deep Cloud Opacity - NH4SH Cloud
Optical Depth of Deep Cloud
8
6
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
4
2
0
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
(c) Upper Cloud Opacity - NH4SH Cloud
Optical Depth of Upper Cloud
2.0
1.5
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
1.0
0.5
0.0
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
Fig. 10. Demonstration of the dependence of retrieved cloud optical depths on the chosen aerosol models (T = non-scattering, S = scattering). (a) and (b) show the opacity of
the deep cloud and the spurious emission angle dependence when grey scatterers are assumed. (c) shows that the north–south asymmetry in upper-cloud opacity is present
for all model assumptions. The corresponding v2 are shown in Fig. 9.
Saturn’s aerosols, as previous reflected sunlight studies of Saturn’s
clouds suggest a latitudinally-uniform s1 or an equator-to-pole decrease in opacity (e.g., Stam et al., 2001; Karkoschka and Tomasko,
2005). The NH4SH scattering cloud produced optical depths for the
deep cloud which were largely independent of latitude (Fig. 10b).
Ultimately the meridional variation of s1 cannot be uniquely determined unless (i) each latitude is viewed with the same emission
angle; or (ii) multiple emission angles are used to separate the
degenerate effects of emission angle and s1. No emission angle
dependence is detected in retrievals of the upper cloud, s2
(Fig. 10c), which shows an asymmetry between northern and
southern hemispheres for all of the cloud models tested (models
II and IV featured the upper cloud), although the retrieved optical
depths are highly dependent on the chosen aerosol model.
4.2.2. Degeneracies in gaseous composition
Unfortunately, the degeneracy between the different cloud
models provides substantial uncertainties in the absolute abundances of gases derived from the 5-lm window. The relative variations of ammonia (shown in Fig. 11a for the grey-cloud optical
model) are similar for all cloud models, but there are clear offsets
in absolute abundance. The use of scattering clouds increases the
pathlength for individual photons in the upper troposphere, and
hence reduces the amount of each gas necessary to reproduce
523
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
(a) NH3 Mole Fraction (ppm)
NH3 Mole Fraction (ppm)
500
400
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
300
200
100
0
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
(b) AsH3 Mole Fraction (ppb)
AsH3 Mole Fraction (ppb)
5.0
4.5
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
4.0
3.5
3.0
2.5
2.0
1.5
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
(c) PH3 Mole Fraction (ppm)
PH3 Mole Fraction (ppm)
4.5
4.0
1 Cloud (T)
1 Cloud (S)
1 Extended (T)
1 Extended (S)
2 Cloud (T)
2 Cloud (S)
Ext/Compact (T)
Ext/Compact (S)
3.5
3.0
2.5
2.0
90
80
70
60
50
40
30
20
10
0
-10 -20 -30 -40 -50 -60 -70 -80 -90
Planetocentric Latitude
Fig. 11. Demonstration of the degeneracy in gaseous distributions of NH3, AsH3 and PH3 depending on the choice of aerosol optical models (T = non-scattering, S = scattering)
and vertical models (I–IV). The corresponding v2 are shown in Fig. 9.
the absorption features (grey curves in Fig. 11 are systematically
lower than the black non-scattering curves). Compact cloud models tend to yield smaller retrieved abundances than extended
clouds, and even the best-fitting aerosol models differ in abundance by a factor of 2–3. Nevertheless, the enhanced NH3 abundances at 45°N, 25°S and the equator are persistent features,
irrespective of the chosen cloud model.
However, in the cases of AsH3 and PH3, the chosen cloud model
can have a substantial effect on both the meridional structure and
the absolute abundances. AsH3 (Fig. 11b) shows a north–south
asymmetry in the non-scattering case that becomes much smaller
when multiple scattering in the southern hemisphere is taken into
account. The distribution of PH3 is even more problematic, with a
large scatter in measured abundances (Fig. 11c), although each aerosol model generally produces a north–south asymmetry in PH3.
The formal retrieval error on each PH3 measurement is small, given
that this gas dominates the shape of the VIMS spectrum, but the
degeneracy between the cloud models makes a determination of
the absolute abundance near-impossible without additional
constraints.
The cause of this offset in absolute abundance is demonstrated
in Fig. 12 for 15°N, 15°S and the equator (using non-scattering
cloud model II.A), which shows how the algorithm converges to
the optimal solution. There is a large variation of retrieved parameters with the cloud base pressure, showing how sensitive the
absolute abundances are to the choice of aerosol model. As we
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
EZ
χ2/N
0.35
0.30
1.0 1.5 2.0 2.5 3.0 3.5
4.5
4.0
1.0 1.5 2.0 2.5 3.0 3.5
4.0
3.8
3.6
3.4
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Base Pressure (bar)
PH3 FSH
PH3 FSH
0.30
0.25
0.20
0.15
0.10
0.05
1.0 1.5 2.0 2.5 3.0 3.5
0.35
0.30
0.25
0.20
0.15
0.10
0.05
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
1.0
0.8
0.6
0.4
1.0 1.5 2.0 2.5 3.0 3.5
1.0
0.9
0.8
0.7
0.6
0.5
0.4
1.0 1.5 2.0 2.5 3.0 3.5
4.0
3.5
3.0
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
PH3 FSH
0.08
0.07
0.06
0.05
0.04
0.03
0.02
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Deep Cloud Opacity
1.2
1.0
0.8
0.6
0.4
0.2
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Upper Cloud Opacity
1.2
1.1
1.0
0.9
0.8
0.7
1.0 1.5 2.0 2.5 3.0 3.5
Upper Cloud τ2
Upper Cloud Opacity
1.2
1.1
1.0
0.9
0.8
0.7
0.6
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Upper Cloud Opacity
1.0
0.8
0.6
0.4
0.2
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Base Pressure (bar)
Base Pressure (bar)
AsH3 Mole Fraction
AsH3 Mole Fraction
AsH3 Mole Fraction
5.1
5.0
4.9
4.8
4.7
4.6
4.5
4.4
1.0 1.5 2.0 2.5 3.0 3.5
Mole Fraction (ppb)
Mole Fraction (ppb)
Upper Cloud τ2
Base Pressure (bar)
4.4
4.3
4.2
4.1
4.0
3.9
1.0 1.5 2.0 2.5 3.0 3.5
NH3 Mole Fraction
160
140
120
100
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
3.6
3.4
3.2
3.0
2.8
2.6
2.4
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Mole Fraction (ppm)
Base Pressure (bar)
180
4.5
Deep Cloud Opacity
Deep Cloud τ1
Deep Cloud τ1
1.2
PH3 Mole Fraction
5.0
Base Pressure (bar)
Deep Cloud Opacity
Mole Fraction (ppm)
PH3 Mole Fraction
Mole Fraction (ppm)
5.0
4.2
Base Pressure (bar)
PH3 FSH
5.5
PH3 FSH
PH3 FSH
Mole Fraction (ppm)
PH3 Mole Fraction
6.0
1.2
1.0
0.8
0.6
0.4
0.2
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Mole Fraction (ppm)
Base Pressure (bar)
Deep Cloud τ1
0.40
15N
Mole Fraction (ppb)
χ2/N
0.45
0.50
0.48
0.46
0.44
0.42
0.40
0.38
1.0 1.5 2.0 2.5 3.0 3.5
χ2/N
15S
550
NH3 Mole Fraction
500
450
400
350
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Base Pressure (bar)
Mole Fraction (ppm)
0.50
Upper Cloud τ2
524
NH3 Mole Fraction
200
180
160
140
120
1.0 1.5 2.0 2.5 3.0 3.5
Base Pressure (bar)
Fig. 12. Trade off between the different atmospheric parameters for three latitudes, 15°S (left column), the equator (central column) and 15°N (right column). The seven rows
show the variation of v2/N with base pressure, where N is the number of spectral channels (N = 32, so a Dv2 = 1 envelope corresponds to 0.03 in these panels) in the retrieval;
the PH3 deep mole fraction and fractional scale height; the opacity of the deep and upper clouds in model II.A; the well-mixed mole fractions of AsH3 and NH3. The vertical
dashed line shows the best-fitting base pressure for each latitude (note that it is poorly constrained in the southern hemisphere case).
525
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
PH3 Mole Fraction (ppm)
AsH3 Mole Fraction (ppb)
5
Retrieved Value
Retrieved Value
3.5
4
3
2
3.0
2.5
2.0
1.5
2
3
4
5
1.5
2.0
2.5
3.0
3.5
True Value
True Value
Deep Cloud Opacity τ1
Upper Cloud Opacity τ2
3.0
5
Retrieved Value
Retrieved Value
2.5
2.0
1.5
1.0
3
2
1
0.5
0.5
1.0
1.5
2.0
1.0
1.5
2.0
2.5
3.0
3.5
True Value
True Value
Base Pressure of Deep Cloud
NH3 Mole Fraction (ppm)
3.0
Retrieved Value
Retrieved Value
4
2.5
2.0
800
600
400
200
1.6
1.8
2.0
2.2
2.4
2.6
2.8
True Value
200
400
600
800
1000
True Value
Retrieved Value
GeH4 Mole Fraction (ppb)
0.5
0.4
0.3
0.20 0.25 0.30 0.35 0.40 0.45 0.50
True Value
Fig. 13. Scatter plots showing positive correlations between synthetic VIMS spectral inputs (‘true’ values) and the retrieved outputs. The only figures showing no correlation
is GeH4, which cannot be reliably retrieved from the VIMS data. The deviation between true and retrieved parameters (the dotted line shows a 1:1 relationship) is used to
define the random error on retrieved quantities.
described above, the deep cloud base pressure for the southern
hemisphere (with the thickest upper cloud) is poorly constrained
and could be placed anywhere at p > 2 bar (first row of Fig. 12).
Placing cloud opacity at greater depth requires larger abundances
526
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
of PH3 (2nd row) and NH3 (7th row), but smaller abundances of
AsH3 (6th row) to reproduce the absorption features. Furthermore,
any variations in the deep cloud opacity s1 are largely compensated by the upper cloud s2 (4th and 5th rows of Fig. 12). This figure demonstrates the tradeoffs between the parameters in
retrievals at each latitude, so we conclude that the absolute abundances and optical depths are dependent on the correct parameterisation of Saturn’s clouds.
4.3. Validation experiments
A robust way of demonstrating the validity of the retrieval
scheme is to attempt extraction of the same variables from modelled VIMS spectra with simulated noise. Two hundred spectra
were synthesised with non-scattering cloud model II.A at a range
of latitudes (±45°) and emission angles (0–45°); a range of a priori
abundances for PH3, NH3, AsH3 and GeH4 (the latter as a control);
and a range of values for the deep cloud s1 and pb and upper cloud
s2. The synthetic spectra were randomised using the same noise
levels described for the real VIMS spectra (Section 2.1), and then
the parameters were simultaneously retrieved from the synthetic
spectra. Fig. 13 shows a positive correlation between the true values and retrieved values for each parameter, with the exception of
GeH4. The average deviations between modelled and retrieved values are: PH3 (8%), AsH3 (7.3%), NH3 (30%), s1 (59%), s2 (25%) and pb
(17.6%). In addition, the NH3 abundance appeared to be 30% lower
than the true values, whereas the retrieved s1 was 54% larger than
the input values. The large uncertainties on s1, s2 and pb demonstrate the high correlation between these parameters, and the difficulties in separating them in the retrievals.
These simple experiments provide estimates of the uncertainties in absolute abundances based solely on random measurement
errors. They do not represent the uncertainties due to systematic
offsets. As we have seen, abundance uncertainties are dominated
by the choice of cloud parameterisation rather than random error
on the VIMS spectra. Relative spatial variability in retrieved quantities are more robust, and these will be presented in Section 5.
5. Results
Atmospheric composition (parameterised PH3; well-mixed NH3
and AsH3) and aerosol properties (s1, s2 and pb using the 2-cloud
scheme, model II) were retrieved from 107 VIMS 4.6–5.1 lm spectra between 38°S and 67°N (planetocentric). The meridional distribution of each parameter is shown in Fig. 14, with the zonal mean
radiances and brightness temperatures at 5 lm indicated in the
top panels and the best-fitting spectral models in Fig. 15. Given
the difficulties in distinguishing between scattering and non-scattering cases with the grey or NH4SH optical properties (models A
and C) based on the v2 alone (Fig. 14b), we applied both techniques
to the VIMS retrievals. Pure NH3 ice and isotropic scattering were
previously ruled out, although we stress that the retrieved atmospheric composition was very similar in these cases. The meridional distribution of v2 (Fig. 14b) shows a small improvement using
multiple scattering with the phase function of NH4SH, but the effect is insignificant within a Dv2 = 1. Although the scattering cloud
is more physically realistic, its inclusion has a substantial effect on
retrieved parameters for such a small improvement in v2, so both
sets of results are shown to highlight the degeneracy issue.
5.1. Saturn’s clouds
The retrieved properties of the compact cloud scheme are
shown in Fig. 14c–e. The base pressure of the deep cloud is poorly
constrained in the southern hemisphere where significant opacity
due to aerosols in the upper cloud (Fig. 14c) and PH3 (Fig. 14g and
h) prevent a unique determination of the deep cloud base. The
equatorial cloud is allowed to be present at lower pressures
(approximately 2.1 bar) in the non-scattering case, compared to
high pressures of the northern hemisphere cloud deck (2.5–
2.8 bar). The need for this 2.1-bar equatorial cloud is removed
when multiple scattering is used, when the equatorial cloud base
becomes consistent with northern mid-latitudes. Both the scattering and non-scattering models agree on the cloud base pressures at
northern mid-latitudes. Seasonally-variable cloud opacities in the
2–3 bar region are deemed unlikely given the long radiative timescales at these pressures, so a cloud base in the 2.5–2.8 bar region
is likely to exist globally on Saturn, with upward advection pushing
the cloud higher at the equator.
Optical depths of the two clouds are higher in the multiple scattering case. The upper cloud s2 (arbitrarily placed at 1.4 bar, representative of the cumulative opacity of clouds and hazes above this
pressure level) is more opaque in the southern hemisphere in both
scattering and non-scattering cases. It is likely that the extended
haze layers between the tropopause and 1.4 bar that are responsible for scattering of reflected sunlight on the dayside (Pérez-Hoyos
et al., 2005) are also contributing to the attenuation of 5-lm flux
on the nightside. Finally, the upper cloud shows enhanced
equatorial opacity only in the multiple-scattering case. Increased
equatorial opacity is qualitatively expected when we consider
the ‘hazy’ appearance of Saturn’s low latitudes in reflected sunlight
(e.g., Porco et al., 2005; Vasavada et al., 2006) and the observations
of vertical upwelling of the disequilibrium species PH3 (Fletcher
et al., 2009a).
The deep cloud opacity (Fig. 14d) shows opposing behaviours
depending on the scattering assumptions. In the scattering case,
we see a trend of increased opacity at high latitudes, whereas the
opposite is true in the non-scattering case. A mean of the two
would be uniform with latitude, which may be more realistic for
the non-seasonal conditions in the 2–3 bar pressure regime.
Small-scale variations in s1 of approximately 20–30% are colocated
in the two cases, but the amplitude of the opacity variation is likely
to depend on the spatial resolution of the VIMS images. Comparing
to Fig. 1, the narrow axisymmetric bands of bright 5-lm flux are
coincident with regions of lower opacity (particularly evident between 20 and 30°N). Fig. 14d suggests that these bright bands
are regions of diminished opacity of the deep cloud layer, rather
than being due to changes in the base pressure of the 2.5–2.8 bar
cloud or the opacity of the upper ‘haze’. Finally, unlike the elevated
opacity of the upper cloud at low latitudes, there it nothing notable
about the deep cloud opacity at the equator.
In summary, VIMS nightside spectra are consistent with clouds
in two regimes – (i) a compact, meridionally-uniform cloud deck
centred in the 2.5–2.8 bar region with small-scale opacity variations responsible for the narrow, bright axisymmetric lanes in
VIMS images; and (ii) a hemispherically-asymmetric upper cloud
above 1.4 bar, whose exact altitude and vertical structure are not
constrained by VIMS, but which is likely to extend towards the tropopause and is responsible for reflected sunlight scattering. The
upper cloud/haze is seasonally variable, whereas the deep cloud
is not. Degeneracy between the scattering and non-scattering cases
leads to uncertainties in absolute optical depths, and elevated
equatorial opacity is only present in the scattering case.
Finally, although a 2.5–2.8 bar cloud deck of NH4SH provided
the best fits to the spectra for the limited range of clouds tested
in this study, this solution is certainly non-unique and we cannot
rule out a more complex combination of NH3, NH4SH and possibly
P2H4 (see Section 3.4). The cloud deck is deeper than the predicted
condensation altitudes for pure NH3 (1.47–1.81 bar, Table 4 of
Atreya et al., 1999, for solar and fivefold enrichments of heavy
elements), but also higher than the predicted levels of NH4SH
527
(a) Radiance at 5 μm
700
600
500
400
300
200
70 60 50 40 30 20 10
0 -10 -20 -30 -40 -50 -60 -70
(f) Brightness Temperature at 5 μm
Brightness Temperature (K)
Radiance (nW/cm2/sr/μm)
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
185
180
175
170
70 60 50 40 30 20 10
Planetocentric Latitude
(b) χ2 of Fits
(g) PH3 Fractional Scale Height
Non-Scattering
Scattering (NH4SH)
0.5
χ2
0.4
0.3
0.2
0.1
70 60 50 40 30 20 10
Fractional Scale Height
0.6
0 -10 -20 -30 -40 -50 -60 -70
0.5
0.4
0.3
0.2
0.1
0.0
70 60 50 40 30 20 10
Planetocentric Latitude
(h) PH3 Mole Fraction (ppm)
(c) Optical Depth of Upper Cloud
Mole Fraction (ppm)
Optical Depth τ2
3.0
2.5
2.0
1.5
1.0
0.5
6
5
4
3
2
70 60 50 40 30 20 10
0 -10 -20 -30 -40 -50 -60 -70
Planetocentric Latitude
(d) Optical Depth of Deep Cloud
(i) NH3 Mole Fraction (ppm)
Mole Fraction (ppm)
Optical Depth τ1
0 -10 -20 -30 -40 -50 -60 -70
Planetocentric Latitude
3.0
2.5
2.0
1.5
1.0
0.5
0.0
70 60 50 40 30 20 10
500
400
300
200
100
0
70 60 50 40 30 20 10
0 -10 -20 -30 -40 -50 -60 -70
0 -10 -20 -30 -40 -50 -60 -70
Planetocentric Latitude
Planetocentric Latitude
(e) Base Pressure of Deep Cloud
(j) AsH3 Mole Fraction (ppb)
2.0
2.2
2.4
84
2.6
2.8
3.0
70 60 50 40 30 20 10
0 -10 -20 -30 -40 -50 -60 -70
Planetocentric Latitude
Mole Fraction (ppb)
1.8
Base Pressure of Deep
Cloud (bar)
0 -10 -20 -30 -40 -50 -60 -70
Planetocentric Latitude
3.5
0.0
70 60 50 40 30 20 10
0 -10 -20 -30 -40 -50 -60 -70
Planetocentric Latitude
5
4
3
2
70 60 50 40 30 20 10
0 -10 -20 -30 -40 -50 -60 -70
Planetocentric Latitude
Fig. 14. Meridional distributions of Saturn’s cloud and aerosol properties (c–e) and gaseous distributions (g–j), for the two best-fitting cloud models: an upper haze and a
deep compact cloud, with the non-scattering grey assumption (solid line, model A) and the scattering NH4SH assumption (dotted line, model C). These are compared to the
zonal mean radiances and brightness temperatures in (a) and (f), respectively. Although the scattering model shows a small improvement in v2 in (a), suggesting that the
optical properties of NH4SH produce the best results, this improvement is deemed insignificant given the degeneracies discussed in the main text. The points with error bars
at 60°S show the formal retrieval uncertainty in each quantity.
(4.56–5.72 bar). Homogeneous cloud condensation occurs when
the partial pressure of a gas exceeds its saturation vapour pressure.
Formation of solid NH4SH is more complex, involving a two-component reaction between NH3 and H2S whose equilibrium can be
expressed by the empirical equation (Lewis and May, 1969; Atreya,
1986);
logðpNH3 pH2 S Þ ¼ 14:82 4705
T
ð1Þ
where pNH3 and pH2 S are the partial pressures of the two gases.
Assuming that the abundances of the two gases are equal at the
cloud condensation altitude (H2S is completely used up in this
528
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
Modelled VIMS Data
Radiance (μW/cm2/sr/μm)
0.8
0.6
30N
15N
EZ
15S
30S
0.4
0.2
0.0
4.6
4.7
4.8
4.9
5.0
5.1
Wavelength (μm)
Fig. 15. The best-fitting spectral models to five selected latitudes. Both the thermal
non-scattering and NH4SH scattering models (lines) produce near-identical fits to
the data (individual points). Overfitting at 4.67 and 4.85 lm, and underfitting at
5.06 lm, are common features of all spectral models and could not be explained by
the addition of further gaseous species.
reaction, whereas NH3 survives to condense at higher, cooler altitudes), and comparing the saturated vapour pressure curve to Saturn’s temperature profile, we require approximately 2 ppm of H2S
to form the VIMS 2.5–2.8 bar cloud. This would be produced by only
10% of the solar S/H ratio of Grevesse et al. (2007), considerably
smaller than the 10 solar S/H abundance suggested by Briggs
and Sackett (1989) (equivalent to 250 ppm), but larger than the
16 ppb upper limit of Weisstein and Serabyn (1996). Simple thermodynamic theory is a poor approximation to Saturn’s true clouds,
given that they are unlikely to be pure ice condensates and probably
contain a range of impurities. The VIMS 2.5–2.8 bar cloud cannot be
identified unambiguously using the present dataset.
5.2. Gaseous composition
Fig. 14g–j shows the meridional distributions of PH3, NH3 and
AsH3 in the scattering and non-scattering cases. In all three cases
scattering increases the path length of individual photons and
hence reduces the abundances required to reproduce the absorption features.
5.2.1. Phosphine
Section 4 demonstrated the uncertainties in the meridional distribution of phosphine under different scattering assumptions. The
fractional scale height (representing the abundance for p < 1.3 bar)
shows a local maximum at the equator under non-scattering conditions, consistent with the distribution identified by Cassini/CIRS
in the 0.1–0.8 bar region (Fig. 7 of Fletcher et al., 2009a). Furthermore, VIMS successfully reproduces the mid-latitude asymmetry
in the fractional scale height, the local minimum at 10–20°S and
the rising abundance towards 40°S observed by CIRS. The asymmetry in the fractional scale height at higher altitudes may be due to
enhanced shielding by southern-hemisphere aerosols, increasing
photolysis lifetimes in the south and allowing PH3 to accumulate
over the summer/autumn season.
But there are problems with these PH3 results: (i) the knee pressure of the distribution (p0 = 1.3 bar) is considerably deeper in the
VIMS retrievals than the CIRS retrievals (p0 = 0.55 bar); (ii) the
deep mole fractions in the scattering (mean and standard error
3.1 ± 0.3 ppm) and non-scattering (4.4 ± 0.6 ppm) cases are smaller
than the 6.4 ± 0.4 ppm mole fraction reported by CIRS (Fletcher
et al., 2009a); (iii) both the scattering and non-scattering cases feature a local minimum in the deep (p > 1.3 bar) equatorial abundance which was not observed by CIRS (CIRS is insensitive to
p > 0.8 bar); and (iv) the need for the equatorial enhancement in
the fractional scale height is removed by the inclusion of scattering. Indeed, on the last point it seems that the PH3 fractional scale
height and the s2 of the upper cloud have exchanged roles in the
retrievals, indicating a degeneracy between the two variables.
The VIMS-derived mole fraction is also smaller than the disk-averaged 4.5–7.5 ppm range reported for Saturn’s deep troposphere by
Burgdorf et al. (2004), Lellouch et al. (2001), Orton et al. (2000), de
Graauw et al. (1997) and Noll and Larson (1990). Finally, an asymmetry in the deep PH3 abundance in the non-scattering case is
deemed unlikely as the vertical mixing processes responsible for
the presence of this disequilibrium species in the upper troposphere are not expected to be seasonally-variable. No deep asymmetries are observed in the multiple-scattering case.
Tests revealed that the use of the CIRS-derived mole fractions
and p0 could not reproduce the VIMS spectrum adequately for
any choice of cloud model, which leaves us with a conundrum –
even though the meridional distributions are largely similar, the
absolute values are quite different from the two instruments. As
the same retrieval model was used in both studies, one possibility
is that the line data for the pentad polyad at 5-lm are inconsistent
with that of the dyad at 9 lm, making direct comparisons difficult.
Indeed, the 5-lm line data are only accurate to the 20–30% level
(Section 3.2), which may explain some of the discrepancy, but
not all of it. Furthermore, just as VIMS retrievals are prone to
PH3 and aerosol degeneracies, CIRS retrievals are prone to T(p)PH3 degeneracies. Finally, the retrieved high-altitude PH3 is determined by the absorption complex at 4.74-lm: if scattered light
within the instrument artificially enhances the flux in this absorption band (see Section 2.1) then we would require less PH3 than expected from CIRS. Further testing of the PH3–aerosol degeneracy
with improved knowledge of the cloud composition, along with
consistent measurement of the PH3 line data, is required to resolve
this issue.
If we take the VIMS-derived PH3 at face value, then some mechanism must be depleting PH3 above the p0 = 1.3-bar level. PH3 is
thought to be well-mixed by vertical diffusion at depth and depleted at higher altitudes due to photolysis to diphosphine (P2H4,
a candidate for Saturn’s haze) and elemental phosphorous. In the
1–3 bar region of VIMS sensitivity (Fig. 6), photochemical models
suggest that production and loss rates are balanced due to recycling of P2H4 to PH3 (J. Moses, personal communication), so depletion would be unexpected. If the PH3 loss at p < 1.3 bar is real, then
it may simply represent adjustment of the vertical profile between
the well-mixed deep profile and the photolysis regime.
5.2.2. Ammonia
Unlike PH3, the meridional distribution of NH3 was similar for
all of the aerosol models tested, even though the absolute abundances vary between scattering and non-scattering cases in
Fig. 14i. Indeed, the largest discrepancy between scattering and
non-scattering is at the equator and mid-southern latitudes, where
the aerosol opacity was at its largest. NH3 is enhanced at the equator between ±5° latitude by three times the northern mid-latitude
abundances. This enhancement is coincident with the narrow region of diffuse brightness in Fig. 1, and with a narrow prograde
jet identified by Garcı́a-Melendo et al. (2010) which exists in addition to the broad equatorial jet. However, the NH3 enhancement is
confined to a much narrower equatorial region (±5°) than the CIRSderived PH3 enhancement (±20°) in the 0.2–0.8 bar region (Fletcher
et al., 2009a).
Smaller enhancements are also notable in axisymmetric bands
at 23–25°S and 42–47° (planetocentric), coinciding with dark lanes
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
at 5.1 lm (Fig. 1). The northern hemisphere NH3 peak exists between opposing zonal jets (prograde at 41°N, retrograde at 49°N),
suggesting upwelling on the poleward side of the prograde jet.
Interestingly, this jet was the location of a meandering lane known
as the ribbon wave, first discovered by Voyager (e.g., Godfrey and
Moore, 1986). The region north of the jet exhibited significant eddy
activity associated with the wave (Godfrey and Moore, 1986), and
appears to be the location of a dark band near 45°N flanked by 5lm bright regions in Fig. 1. The northward gradient of potential
vorticity (PV) was found to change sign at 44°N near to this jet,
potentially violating the stability criterion of Arnol’d’s second theorem (Read et al., 2009) and suggesting that the eddy activity (and
possibly the enhanced NH3 detected by VIMS) arises due to instabilities in the flow at depth. If the two hemispheres are symmetric
at depth, we might expect a similar NH3 enhancement at southern
mid-latitudes (44–51°S), and indeed Cassini imaging shows wavelike activity and an abundance of small vortices at this latitude
(Vasavada et al., 2006; Choi et al., 2009). Unfortunately, these
southern latitudes were not covered by the nightside VIMS spectra
studied here.
The band at 23–25°S, which is embedded in the region of prograde flow associated with the equatorial jet, is also associated
with a dark band in Fig. 1. The upwelling band is poleward of the
warm South Equatorial Belt (SEB) at 14–17°S, and further north
than Saturn’s ‘storm alley’ (a region between 33 and 40°S characterised by an abundance of vortices, Vasavada et al., 2006), but
may be associated with wave-like activity and tilted streaks observed in the same latitude band (Vasavada et al., 2006; Choi
et al., 2009). Finally, despite these three regions of upwelling, we
cannot unambiguously identify the sink regions of gaseous NH3 required for continuity. However, depletion of gaseous NH3 could be
provided by (i) subsidence in regions flanking the upwelling, (ii)
condensation to form fresh NH3 clouds and (iii) photolysis to form
hydrazine (a possible constituent of Saturn’s tropospheric hazes).
Aside from these three regions of upwelling, the NH3 abundance
is reasonably uniform, varying between 120 and 180 ppm in the
northern hemisphere, and slightly larger (120–220 ppm) in the
south, depending on the scattering assumptions. Given the range
of the results in Fig. 11, the NH3 mole fraction derived from VIMS
is uncertain by a factor of 2. For the best fitting cloud models we
find globally-averaged abundances of 140 ± 50 ppm (scattering)
and 200 ± 80 ppm (non-scattering) in the 1–4 bar sensitivity range
of Fig. 6. The retrieved NH3 abundance can be compared to the partial pressure for 100% relative humidity to estimate the condensation altitudes for the gas. Equatorial NH3 (500 ppm) would
condense near 1.65 bar, whereas the global mean abundance
(140 ppm) suggests condensation at 1.35 bar. This implies that
NH3 is saturated and well-mixed by diffusive processes up to the
1.35–1.65 bar level (consistent with the expected altitude of NH3
condensation, Atreya et al., 1999), and then declines following a
saturated vapour–pressure curve and photolysis at lower pressures. Compared to some of the previous disk-averaged NH3 determinations in Table 1, we find consistency with the 70–120 ppm
values of Briggs and Sackett (1989), Grossman et al. (1989), de Graauw et al. (1997, quoted for the 1.2-bar level), Orton et al. (2000)
and Burgdorf et al. (2004). The VIMS result is within the range of
50–200 ppm measured by Voyager/IRIS (Courtin et al., 1984) and
slightly smaller than the 500 ppm abundance at p > 3 bar derived
from microwave spectra (de Pater and Massie, 1985), except in
the region of strong upwelling at the equator.
5.2.3. Arsine
AsH3 is the principal arsenic-bearing gas on Jupiter and Saturn,
though previous studies have focussed solely on globally-averaged
values. The meridional distribution of AsH3 is shown in Fig. 14j for
the first time. Both scattering and non-scattering cases indicate
529
local maxima flanking the equatorial region, centred on 7°N and
7°S. The two maxima are much closer to the equator than the
warm tropospheric belts (±15°) observed by CIRS (Fletcher et al.,
2007b). However, the non-scattering case predicts an AsH3 asymmetry (from around 4 ppb in the south to 2.5–3.0 ppb in the north)
that is not apparent in the scattering case (uniform abundance of
2.2 ± 0.3 ppb in both hemispheres). The global mean abundances
of AsH3 in the scattering (2.2 ± 0.3 ppb) and non-scattering
(3.3 ± 0.8 ppb) cases are consistent with ground-based measurements of 3.0 ± 1.0 ppb (Noll and Larson, 1990) and 2:4þ1:4
1:2 ppb
(Bézard et al., 1989), although VIMS spectra do not have the spectral resolution to confirm the decreasing abundance with altitude
(presumably due to photolysis) detected by Bézard et al. (1989).
Like PH3, AsH3 can be thought of as a tracer of tropospheric mixing, as its abundance at the altitudes studied by VIMS greatly exceeds thermochemical equilibrium predictions (e.g., Fegley and
Lewis, 1979; Fegley and Lodders, 1994). This disequilibrium is
thought to be caused by vertical transport, mixing parcels of air
from the deep troposphere at a faster rate than AsH3 can be chemically destroyed (conversion to solid phase As4 or As2S2), thus the
tropospheric AsH3 abundance represents Saturn’s equilibrium
composition at much deeper levels (temperatures exceeding
400 K, Fegley and Lodders, 1994). Using the solar photospheric
composition of Grevesse et al. (2007), we estimate a supersolar
As/H ratio of 6.4–9.6 times solar (depending on the scattering
and non-scattering assumptions), larger than the subsolar (0.6)
abundance on Jupiter (Noll et al., 1990), whereas P/H is supersolar
on both planets (Fletcher et al., 2009a). As pointed out by Fegley
and Lodders (1994), this difference is hard to explain because P
and As exhibit similar cosmochemical behaviours, so we might expect equal enrichments of both elements during accretion.
6. Discussion: Possible dynamical mechanisms
While detailed dynamical modelling is deferred to future studies, here we discuss some plausible speculations concerning the
dynamical processes responsible for the retrieved gaseous abundances and cloud distributions in Section 5. Fig. 14 indicated that
the best-fitting VIMS models produce deep PH3 (p > 1.3 bar) and
AsH3 distributions that do not show the same meridional variations as NH3 and high-altitude PH3 (p < 1.3 bar). In particular, deep
PH3 and AsH3 showed local maxima either side of the equator,
whereas the PH3 scale height, the upper cloud opacity and NH3
show maxima directly at the equator. At first glance this is difficult
to interpret in terms of vertical transport from the deep troposphere, but the two different regimes may be reconciled if we consider a scenario where two stacked meridional circulation cells
exist in Saturn’s troposphere (see descriptions by, Del Genio
et al., 2009; Ingersoll et al., 2000; Showman et al., 2005).
Cloud-tracking observations of eddy-momentum convergence
on both Jupiter and Saturn have long indicated that eddies accelerate the jets at pressures of 1 bar or deeper (Ingersoll et al., 1981;
Salyk et al., 2006; Del Genio et al., 2007). In steady state, these eddy
accelerations would be balanced by meridional flow that is equatorward across eastward jets and poleward across westward jets.
This meridional flow also helped to explain the prevalence of thunderstorms in jovian belts (Gierasch et al., 2000; Ingersoll et al.,
2000) and the distribution of NH3 from radio observations (e.g.,
Fig. 3 of Showman et al., 2005). However, these observations need
to be reconciled with the ‘classical’ view of the belt/zone circulation on giant planets, whereby air rises in low-temperature anticyclonic zones on the equatorial flanks of eastward jets and sinks in
warmer cyclonic belts (e.g., Hess and Panofsky, 1951). The resulting meridional circulation causes poleward motion across eastward jets and equatorward motion across westward jets,
530
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
opposing the flow suggested by the jet-pumping scenario described above. In steady state, the zonal Coriolis accelerations implied by this ‘upper cell’ circulation are balanced by an unidentified
source of atmospheric ‘drag’ that decelerates the jets in the upper
troposphere (Conrath and Pirraglia, 1983; Gierasch et al., 1986;
Conrath et al., 1990).
The VIMS results require both the jet-pumping and the jetdamping circulation regimes to be invoked (the stacked-cell
hypothesis). In this scenario, we suggest that drag within 20–30°
of the equator enables a ‘classical’ meridional circulation in the
upper cell. Air rises and diverges (cools) within the equatorial zone,
advecting PH3- and NH3-rich air (along with aerosols to act as
cloud nucleation sites) from depths below the NH3 cloud into the
upper troposphere to explain the equatorial maxima in Fig. 14c,
g and i. This upper-cell air then moves poleward to 10–20°, where
it descends and warms over the equatorial belts , leading to the relatively PH3- and NH3-depleted air at those latitudes. However, this
classical upper-cell circulation must give way in the deeper troposphere to a circulation in the opposite sense. VIMS observations of
off-equatorial maxima (±10°) in AsH3 and deep PH3 (Fig. 14h and
j) suggest that air rises in the belts at 10–20°, moves equatorward
and descends at the equator.
While we stress that the stacked-cell hypothesis may not be a
unique explanation (and further predictive modelling is required),
we note that these two different circulation regimes emerge quite
naturally from considerations of momentum balance of the jets –
the jet-pumping eddies on Jupiter and Saturn likely result from
baroclinic instabilities or moist convection in the adiabatic region
of the deep troposphere. However, convection and instabilities
are largely inhibited in the stably-stratified upper troposphere so
that eddies are confined to the deeper cell, leading to jet damping
(and the opposite sense of meridional circulation) in the upper cell.
The transition between the regimes of differing eddy behaviour
(jet-pumping to jet-damping) may be set by the thermal stratification of the atmosphere, which grows larger in the upper troposphere. Numerical models of jet formation on the giant planets
indeed show deep circulation cells whose tops close at 1 bar,
although they do not consistently capture the hypothesized upper
cells and the jet decay with altitude, perhaps because the appropriate small-scale drag processes in the upper troposphere (e.g.,
absorption of small-scale gravity waves) are not represented (Lian
et al., 2008). The mean circulation of the stacked cells would not be
closed systems, as turbulent small-scale eddy transport would permit mixing of gases (e.g., mean flux of PH3, NH3 and AsH3) and
aerosols vertically between the cells, as well as creating temporal
variability on the cell structure itself. Furthermore, the PH3 and
AsH3 off-equatorial maxima have no counterparts in the zonal jet
structure (which shows a broad prograde jet at the equator), but
small-scale variations in the jet velocity (e.g., those recently detected by Garcı́a-Melendo et al. (2010)) may produce localised vorticity-mixing barriers that could be correlated with the distinct,
narrow cloud lanes.
A second plausible explanation for the VIMS results involves
eddy mixing, which could play an important role in transport of
heat and gaseous species as they do on Earth (e.g., the Ferrel cell,
where eddy heat transport dominates over mean transport). The
mean circulation would produce cold equatorial temperatures on
isobars in the upper cell (upwelling and divergence, as detected
by Cassini/CIRS, Fletcher et al., 2007b) and warm temperatures in
the deep cell (convergence and subsidence). A similar temperature
pattern can also result from a single circulation cell in the presence
of latent heating warming the atmosphere at depth (e.g., Fig. 9 of
Lian et al., 2010). Because the air is statically stable, isentropes (surfaces of constant entropy) would bow upward at the equator in the
upper cell and downward in the lower cell (e.g., Fig. 5 of Showman
et al., 1998). Mixing by eddy transport is almost isentropic on
Saturn because of the long radiative time constant. Hence eddy
mixing in the upper cell transports NH3 and PH3-laden air upward
and equatorward from greater pressures (off the equator) to lower
pressures (at the equator). Similarly, eddy mixing in the deep cell
would transport PH3 and AsH3-poor air downward and equatorward from lower pressures (off the equator) to greater pressures
(at the equator). Thus quasi-isentropic mixing by eddies could occur simultaneously with a mean non-isentropic meridional flow
(air crosses isentropes as it is heated and cooled), and both processes are capable of explaining the meridional distributions of
PH3, AsH3 and NH3 observed by VIMS.
Unfortunately, the spectral resolution of the VIMS data is too
low to permit full 2D (i.e., latitude and altitude) retrievals of PH3,
AsH3 and NH3 which would allow further study of these different
regimes. Furthermore, Section 4 indicated that the separation of
gaseous composition and aerosol scattering/absorption is certainly
non-unique. Although sensitivity extends over the 1–4 bar range in
Fig. 6, we can obtain only a single 1D (i.e., latitudinal) estimate for
AsH3 and NH3 abundances. Although the 1D distributions of these
gases are consistent with the stacked-cell hypothesis, 2D distributions from high spectral-resolution mapping of these dynamical
tracers is required to make advances in this field. However, this
hypothesis may also explain why visible reflectivity (which exhibits albedo contrasts characteristic of the upper-cell meridional circulation in the jet-damping region) appears so different from
Saturn’s 5-lm appearance in Fig. 1 (representing the jet pumping
in the deeper meridional cell).
7. Conclusions
Cassini/VIMS maps of Saturn’s 4.6–5.1 lm nightside thermal
emission have been used to study the latitudinal distribution of
opacity sources in Saturn’s troposphere between 38°S and 67°N
(planetocentric). The spatial variation of atmospheric composition
(PH3, NH3 and AsH3) and aerosol properties (the opacities of a compact 2.5–2.8 bar cloud and aerosols at p < 1.4 bar) are used to probe
the vertical dynamics and chemistry in the NH3 and NH4SH ice
cloud-forming regions of Saturn’s troposphere. The spatial variability of Saturn’s NH3 and AsH3 have been measured for the first time.
Although the parameterisation of the aerosol model (scattering
versus non-scattering; compact versus extended clouds; size distribution and refractive indices) has a significant effect on the retrieved opacities and gaseous abundances, we find that relative
spatial variability can be retrieved reliably from the VIMS spectra
even if absolute abundances remain uncertain. This study provides
the following conclusions:
1. VIMS sensitivity: Maps of Saturn’s thermal emission at 4.6–
5.1 lm reveal a previously unseen dynamical regime in the adiabatic region of the troposphere, with numerous narrow lanes of
opacity variations (particularly the dark lane ±5° of the equator);
a strong mid-latitude seasonal asymmetry in emission between
±5° and ±32°; and a plethora of discrete cloud features. This deep
regime may be the region of eddy convergence which supplies
momentum to the prograde jets (e.g., Del Genio et al., 2009), below
the jet-drag region of the thermally-stratified upper troposphere.
However, VIMS spectra are also sensitive to upper tropospheric
clouds/hazes, with a seasonally-generated asymmetry in opacity
attenuating the thermal emission. Extensive testing of the retrieval
model indicated VIMS sensitivity to both atmospheric composition
(parameterised PH3, well-mixed NH3 and AsH3, but not GeH4, CO,
H2O or CH4) and cloud properties.
2. Saturn’s clouds: Spectral fitting was consistent with cloud
opacity in two regimes – (i) a compact, meridionally-uniform cloud
deck centred in the 2.5–2.8 bar region with small-scale opacity
variations (20–30% at the resolution of the VIMS images used in
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
this study) responsible for the narrow, bright axisymmetric lanes
in VIMS images; and (ii) a hemispherically asymmetric upper cloud
above the 1.4-bar level, whose exact altitude and vertical structure
are not constrained by VIMS, but which is likely to extend towards
the tropopause and is responsible for reflected sunlight scattering
on the dayside. The upper cloud shows a 1.5–2.0 times enhanced
opacity within ±10° of the equator. A scheme with a single-cloud
layer was indistinguishable from the 2-cloud scheme at northern
mid-latitudes, where the opacity of the upper cloud is at its smallest. The deep cloud base is poorly constrained in the southern
hemisphere (it must exist at p > 2 bar) where the opacity of the
upper cloud is at its largest. The meridional opacity distribution
is highly sensitive to the optical properties of the clouds, but of
the limited range of cloud compositions tested here, the optical
constants of NH4SH provided the best fits to the VIMS spectra.
The deep cloud is not likely to consist of pure NH3 ice, but more
complex cloud compositions (e.g., a mixture of NH3 and NH4SH;
or the presence of P2H4 and other contaminants) cannot be ruled
out. The 2.5–2.8 bar cloud is deeper than the predicted condensation altitude of NH3 (1.81 bar for a 5 enrichment of heavy elements, Atreya et al., 1999) and higher than the predicted levels
for NH4SH condensation (5.72 bar), so its composition cannot be
identified unambiguously.
3. Phosphine: PH3 dominates the morphology of the 5-lm spectrum, but its meridional variation is highly sensitive to the choice
of cloud model. A well-mixed PH3 distribution failed to reproduce
the spectrum, and we found that the abundance begins to decline
for p < 1.3 bar (lower pressures at the equator). The fractional scale
height for the upper-tropospheric PH3 generally showed a maximum at the equator and a mid-latitude asymmetry (consistent
with the results from Cassini/CIRS, Fletcher et al., 2009a). The deep
PH3 showed an equatorial minimum flanked by two off-equatorial
maxima (±10°). However, deep mole fractions in the scattering
(mean and standard error 3.1 ± 0.3 ppm) and non-scattering
(4.4 ± 0.6 ppm) cases were smaller than the 6.4 ± 0.4 ppm mole
fraction reported by CIRS, and the p0 = 1.3-bar transition from the
well-mixed to the photolysis region was much deeper than that
derived from CIRS (p0 = 0.55 bar). Uncertainties in the cloud spectral properties, as well as the PH3 line data, are the likely source
of this CIRS–VIMS discrepancy, requiring joint modelling to resolve
this issue.
4. Ammonia: NH3 has a significant effect on the spectrum near
5.1 lm and a similar spatial distribution for all cloud models
tested, being elevated within ±5° of the equator (in a region of
strong 5-lm attenuation) by three times the northern mid-latitude
abundances. Extratropical upwelling is also suggested by small
enhancements at 23–25°S and 42–47°N. The northern peak is associated with a 5-lm dark band just north of the prograde jet at
41°N, and may be associated with abundant eddy activity and
the ‘ribbon wave’ at this latitude. Aside from the three regions of
upwelling, the NH3 abundance was latitudinally uniform, with
globally averaged 1–3 bar abundances of 140 ± 50 ppm (scattering)
and 200 ± 80 ppm (non-scattering), rising to 300–500 ppm at the
equator.
5. Arsine: The spatial variability of Saturn’s principal arsenicbearing gas has been measured for the first time, showing local
maxima at ±7° and a minimum at the equator. An AsH3 asymmetry
(from around 4 ppb in the south to 2.5–3.0 ppb in the north) was
detected using non-scattering clouds, but is not apparent in the
more physically-realistic scattering models (uniform abundance
of 2.2 ± 0.3 ppb in both hemispheres). This results in a supersolar
As/H ratio of 6.4–9.6 times solar, larger than the subsolar (0.6)
abundance on Jupiter. This difference between the two gas giants
is unexpected, as P/H is supersolar on both planets and the two
species should have shared many common properties during planetary accretion.
531
Exploitation of the 5-lm window by Cassini/VIMS has revealed
a planet with symmetric dynamics at depth coupled to substantial
seasonal asymmetries in the upper troposphere. However, uncertainties in the properties and distribution of Saturn’s clouds produces significant degeneracies in modelling the VIMS data.
Future work should focus on (a) comparing dayside 4.6–5.1 lm
spectra to those on the nightside to quantify the effects of sunlight
scattering; (b) exploiting 1–4 lm reflection spectroscopy of Saturn’s clouds to constrain the vertical aerosol distribution and phase
function; (c) incorporating new constraints on aerosol size distributions and optical properties to constrain gaseous retrievals;
and (d) producing regional maps of isolated dynamic features to
qualitatively assess the physical reality of the retrieval model. Future near-infrared instruments for giant planet exploration should
feature improved spectral resolutions in the 5-lm window to
break the degeneracies between aerosols and composition and permit fully three-dimensional retrievals to trace tropospheric
dynamics within and beneath the condensation clouds.
Acknowledgments
Fletcher was supported during this research by a Glasstone Science Fellowship at the University of Oxford. Irwin acknowledges
the support of the UK Science and Technology Facilities Council.
Orton carried out part of this research at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with
NASA, and acknowledges support from the Cassini Project. We
thank the members of the VIMS investigation team who have assisted in the design of the imaging sequences, instrument commands and other vital operational tasks, and the Ground Systems
Operations for the Cassini Project. This research has made use of
the USGS Integrated Software for Imagers and Spectrometers (ISIS).
References
Atreya, S.K., 1986. Atmospheres and ionospheres of the outer planets and their
satellites. Physics and Chemistry in Space, vol. 15. Springer-Verlag, Berlin
Heidelberg.
Atreya, S.K., Wong, M.H., Owen, T.C., Mahaffy, P.R., Niemann, H.B., de Pater, I.,
Drossart, P., Encrenaz, T., 1999. A comparison of the atmospheres of Jupiter and
Saturn: Deep atmospheric composition, cloud structure, vertical mixing, and
origin. Planet. Space Sci. 47, 1243–1262.
Baines, K.H., Drossart, P., Momary, T.W., Formisano, V., Griffith, C., Bellucci, G.,
Bibring, J.P., Brown, R.H., Buratti, B.J., Capaccioni, F., Cerroni, P., Clark, R.N.,
Coradini, A., Combes, M., Cruikshank, D.P., Jaumann, R., Langevin, Y., Matson,
D.L., McCord, T.B., Mennella, V., Nelson, R.M., Nicholson, P.D., Sicardy, B., Sotin,
C., 2006. The atmospheres of Saturn and Titan in the near-infrared: First results
of Cassini/VIMS. Earth, Moon, Planets, 1–29.
Baines, K.H., Momary, T.W., Roos-Serote, M., The Cassini/VIMS Science Team, 2006.
North vs south on Saturn: Discovery of a pronounced hemispherical asymmetry
in Saturn’s 5-micron emission and associated deep cloud structure by Cassini/
VIMS. Bull. Am. Astron. Soc. 38, 488.
Baines, K.H., Momary, T.W., Fletcher, L.N., Showman, A.P., Roos-Serote, M., Brown,
R.H., Buratti, B.J., Clark, R.N., Nicholson, P.D., 2009. Saturn’s north polar cyclone
and hexagon at depth revealed by Cassini/VIMS. Planet. Space Sci. 57, 1671–
1681.
Bevington, P.R., Robinson, D.K., 1992. Data Reduction and Error Analysis for the
Physical Sciences, second ed. WBC/McGraw-Hill, New York.
Bézard, B., Baluteau, J.P., Marten, A., 1983. Study of the deep cloud structure in the
equatorial region of Jupiter from Voyager infrared and visible data. Icarus 54,
434–455.
Bézard, B., Drossart, P., Lellouch, E., Tarrago, G., Maillard, J.P., 1989. Detection of
arsine in Saturn. Astrophys. J. 346, 509–513.
Borysow, A., 1991. Modelling of collision-induced infrared-absorption spectra of
H2–H2 pairs in the fundamental band at temperatures from 20 K to 300 K. Icarus
92, 273–279.
Borysow, A., 1993. Erratum: Modelling of collision-induced infrared-absorption
spectra of H2–H2 pairs in the fundamental band at temperatures from 20 to
300 K [Icarus, Vol. 92, No.2, p. 273–279 (Aug. 1991)]. Icarus 106, 614.
Borysow, A., Frommhold, L., 1986. Theoretical collision-induced rototranslational
absorption spectra for the outer planets: H2–CH4 pairs. Astrophys. J. 304, 849–
865.
Borysow, A., Frommhold, L., 1987. Collision-induced rototranslational absorption
spectra of CH4–CH4 pairs at temperatures from 50 to 300 K. Astrophys. J. 318,
940–943.
532
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
Borysow, J., Frommhold, L., Birnbaum, G., 1988. Collison-induced rototranslational
absorption spectra of H2–He pairs at temperatures from 40 to 3000 K.
Astrophys. J. 326, 509–515.
Briggs, F.H., Sackett, P.D., 1989. Radio observations of Saturn as a probe of its
atmosphere and cloud structure. Icarus 80, 77–103.
Brown, L., Sams, R., Kleiner, I., Cottaz, C., Sagui, L., 2002. Line intensities of the
phosphine dyad at 10 lm. J. Mol. Spectrosc. 215 (2), 178–203.
Brown, R.H., Baines, K.H., Bellucci, G., Bibring, J.-P., Buratti, B.J., Capaccioni, F.,
Cerroni, P., Clark, R.N., Coradini, A., Cruikshank, D.P., Drossart, P., Formisano, V.,
Jaumann, R., Langevin, Y., Matson, D.L., McCord, T.B., Mennella, V., Miller, E.,
Nelson, R.M., Nicholson, P.D., Sicardy, B., Sotin, C., 2004. The Cassini Visual and
Infrared Mapping Spectrometer (VIMS) investigation. Space Sci. Rev. 115, 111–
168.
Burgdorf, M.J., Orton, G.S., Encrenaz, T., Davis, G.R., Lellouch, E., Sidher, S.D.,
Swinyard, B.M., 2004. Far-infrared spectroscopy of the giant planets:
Measurements of ammonia and phosphine at Jupiter and Saturn and the
continuum of Neptune. Adv. Space Res. 34, 2247–2250.
Butler, R., Sagui, L., Kleiner, I., Brown, L., 2006. The absorption spectrum of
phosphine (ph3) between 2.8 and 3.7 lm: Line positions, intensities, and
assignments. J. Mol. Spectrosc. 238 (2), 178–192.
Choi, D.S., Showman, A.P., Brown, R.H., 2009. Cloud features and zonal wind
measurements of Saturn’s atmosphere as observed by Cassini/VIMS. J. Geophys.
Res. 114, E04007, doi:10.1029/2008JE003254.
Clapp, M.L., Miller, R.E., 1996. Complex refractive indices of crystalline hydrazine
from aerosol extinction spectra. Icarus 123, 396–403.
Conrath, B.J., Gautier, D., 2000. Saturn helium abundance: A reanalysis of Voyager
measurements. Icarus 144, 124–134.
Conrath, B.J., Pirraglia, J.A., 1983. Thermal structure of Saturn from Voyager infrared
measurements – Implications for atmospheric dynamics. Icarus 53, 286–292.
Conrath, B.J., Gierasch, P.J., Leroy, S.S., 1990. Temperature and circulation in the
stratosphere of the outer planets. Icarus 83, 255–281.
Courtin, R., Gautier, D., Marten, A., Bézard, B., Hanel, R., 1984. The composition of
Saturn’s atmosphere at northern temperate latitudes from Voyager IRIS spectra
– NH3, PH3, C2H2, C2H6, CH3D, CH4, and the saturnian D/H isotopic ratio.
Astrophys. J. 287, 899–916.
Cuzzi, J.N., Clark, R.N., Filacchione, G., French, R., Johnson, R., Marouf, E.A., Spilker,
L.J., 2009. Ring particle composition and size distribution. In: Saturn from
Cassini-Huygens. Springer, pp. 459–509 (Chapter 15).
Dana, V., Mandin, J.-Y., Tarrago, G.W.B.O., Drossart, P., 1993. Absolute infrared
intensities in the fundamentals m1 and m3 of arsine. J. Mol. Spectrosc. 159 (22),
468–480.
de Graauw, T., Feuchtgruber, H., Bézard, B., Drossart, P., Encrenaz, T., Beintema, D.A.,
Griffin, M., Heras, A., Kessler, M., Leech, K., Lellouch, E., Morris, P., Roelfsema,
P.R., Roos-Serote, M., Salama, A., Vandenbussche, B., Valentijn, E.A., Davis, G.R.,
Naylor, D.A., 1997. First results of ISO-SWS observations of Saturn: Detection of
CO2, CH3C2H, C4H2 and tropospheric H2O. Astron. Astrophys. 321, L13–L16.
Del Genio, A.D., Barbara, J.M., Ferrier, J., Ingersoll, A.P., West, R.A., Vasavada, A.R.,
Spitale, J., Porco, C.C., 2007. Saturn eddy momentum fluxes and convection:
First estimates from Cassini images. Icarus 189, 479–492.
Del Genio, A.D., Achterberg, R.K., Baines, K.H., Flasar, F.M., Read, P.L., SánchezLavega, A., Showman, A.P., 2009. Saturn atmospheric structure and dynamics.
In: Saturn from Cassini-Huygens. Springer, pp. 113–159 (Chapter 6).
de Pater, I., Massie, S.T., 1985. Models of the millimeter–centimeter spectra of the
giant planets. Icarus 62, 143–171.
Fegley, B., Lewis, J.S., 1979. Thermodynamics of selected trace elements in the
jovian atmosphere. Icarus 38, 166–179.
Fegley, B.J., Lodders, K., 1994. Chemical models of the deep atmospheres of Jupiter
and Saturn. Icarus 110, 117–154.
Ferraro, J.R., Sill, G., Fink, U., 1980. Infrared intensity measurements of
cryodeposited thin films of NH3, NH4HS, H2S, and assignments of absorption
bands. Appl. Spectrosc. 34, 525–533.
Fink, U., Larson, H.P., 1978. Deuterated methane observed on Saturn. Science 201,
343–345.
Fink, U., Larson, H.P., Bjoraker, G.L., Johnson, J.R., 1983. The NH3 spectrum in Saturn’s
5 micron window. Astrophys. J. 268, 880–888.
Fletcher, L.N., Irwin, P.G.J., Teanby, N.A., Orton, G.S., Parrish, P.D., Calcutt, S.B.,
Bowles, N., de Kok, R., Howett, C., Taylor, F.W., 2007a. The meridional phosphine
distribution in Saturn’s upper troposphere from Cassini/CIRS observations.
Icarus 188, 72–88.
Fletcher, L.N., Irwin, P.G.J., Teanby, N.A., Orton, G.S., Parrish, P.D., Calcutt, S.B., de
Kok, R., Howett, C., Bowles, N., Taylor, F.W., 2007b. Characterising Saturn’s
vertical temperature structure from Cassini/CIRS. Icarus 189, 457–478.
Fletcher, L.N., Orton, G.S., Teanby, N.A., Irwin, P.G.J., 2009a. Phosphine on Jupiter and
Saturn from Cassini/CIRS. Icarus 202, 543–564.
Fletcher, L.N., Orton, G.S., Teanby, N.A., Irwin, P.G.J., Bjoraker, G.L., 2009b. Methane
and its isotopologues on Saturn from Cassini/CIRS observations. Icarus 199,
351–367.
Fletcher, L.N., Achterberg, R.K., Greathouse, T.K., Orton, G.S., Conrath, B.J., SimonMiller, A.A., Teanby, N., Guerlet, S., Irwin, P.G.J., Flasar, F.M., 2010. Seasonal change
on Saturn from Cassini/CIRS observations, 2004–2009. Icarus 208, 337–352.
Fusina, L., Di Lonardo, G., 2000. The m2 and m4 bending fundamentals of phosphine
(ph3). J. Mol. Struct. 517, 67–78.
Gaddis, L., Anderson, J., Becker, K., Becker, T., Cook, D., Edwards, K., Eliason, E., Hare,
T., Kieffer, H., Lee, E.M., Mathews, J., Soderblom, L., Sucharski, T., Torson, J.,
McEwen, A., Robinson, M., 1997. An overview of the integrated software for
imaging spectrometers (ISIS). Lunar Planet. Sci. 28, 387–+.
Garcı́a-Melendo, E., Sánchez-Lavega, A., Legarreta, J., Perez-Hoyos, S., Hueso, R.,
2010. A strong high altitude narrow jet detected at Saturn’s equator. Geophys.
Res. Lett. 37, L22204.
Gierasch, P.J., Magalhaes, J.A., Conrath, B.J., 1986. Zonal mean properties of Jupiter’s
upper troposphere from Voyager infrared observations. Icarus 67, 456–483.
Gierasch, P.J., Ingersoll, A.P., Banfield, D., Ewald, S.P., Helfenstein, P., Simon-Miller,
A., Vasavada, A., Breneman, H.H., Senske, D.A., Galileo Imaging Team, 2000.
Observation of moist convection in Jupiter’s atmosphere. Nature 403, 628–630.
Godfrey, D.A., Moore, V., 1986. The saturnian ribbon feature – A baroclinically
unstable model. Icarus 68, 313–343.
Goody, R., West, R., Chen, L., Crisp, D., 1989. The correlated-k method for radiation
calculations in nonhomogeneous atmospheres. J. Quant. Spectrosc. Radiat.
Trans. 42, 539–550.
Grevesse, N., Asplund, M., Sauval, A., 2007. The solar chemical composition. Space
Sci. Rev. 130 (1), 105–114.
Grossman, A.W., Muhleman, D.O., Berge, G.L., 1989. High-resolution microwave
images of Saturn. Science 245, 1211–1215.
Hansen, J.E., Travis, L.D., 1974. Light scattering in planetary atmospheres. Space Sci.
Rev. 16, 527–610.
Hess, S.L., Panofsky, H.A., 1951. The atmospheres of the other planets. In: Compendium
of Meteorology. American Meteorological Society, Boston, pp. 391–398.
Howett, C.J.A., Carlson, R.W., Irwin, P.G.J., Calcutt, S.B., 2007. Optical constants of
ammonium hydrosulfide ice and ammonia ice. J. Opt. Soc. Am. B Opt. Phys. 24,
126–136.
Husson, N., Bonnet, B., Scott, N., Chedin, A., 1992. Management and study of
spectroscopic information – the GEISA program. J. Quant. Spectrosc. Radiat.
Trans. 48, 509–518.
Ingersoll, A., Beebe, R., Mitchell, J., Garneau, G., Yagi, G., Muller, J., 1981. Interaction
of eddies and mean zonal flow on Jupiter as inferred from Voyager 1 and 2
images. J. Geophys. Res. 86, 8733–8743.
Ingersoll, A.P., Gierasch, P.J., Banfield, D., Vasavada, A.R., Galileo Imaging Team,
2000. Moist convection as an energy source for the large-scale motions in
Jupiter’s atmosphere. Nature 403, 630–632.
Irwin, P.G.J., Dyudina, U., 2002. The retrieval of cloud structure maps in the
equatorial region of Jupiter using a principal component analysis of Galileo/
NIMS data. Icarus 156, 52–63.
Irwin, P.J.G., Calcutt, S.B., Taylor, F.W., 1997. Radiative transfer models for Galileo
NIMS studies of the atmosphere of Jupiter. Adv. Space Res. 19, 1149–1158.
Irwin, P.G.J., Weir, A.L., Smith, S.E., Taylor, F.W., Lambert, A.L., Calcutt, S.B., CameronSmith, P.J., Carlson, R.W., Baines, K., Orton, G.S., Drossart, P., Encrenaz, T., RoosSerote, M., 1998. Cloud structure and atmospheric composition of Jupiter
retrieved from Galileo near-infrared mapping spectrometer real-time spectra. J.
Geophys. Res. 12, 23001–23022.
Irwin, P.G.J., Weir, A.L., Taylor, F.W., Calcutt, S.B., Carlson, R.W., 2001. The origin of
belt/zone contrasts in the atmosphere of Jupiter and their correlation with 5lm opacity. Icarus 149, 397–415.
Irwin, P., Teanby, N., de Kok, R., Fletcher, L., Howett, C., Tsang, C., Wilson, C., Calcutt,
S., Nixon, C., Parrish, P., 2008. The NEMESIS planetary atmosphere radiative
transfer and retrieval tool. J. Quant. Spectrosc. Radiat. Trans. 109 (6), 1136–1150.
Jacquinet-Husson, N., Scott, N.A., Chedin, A., Garceran, K., Armante, R., Chursin, A.A.,
Barbe, A., Birk, M., Brown, L.R., Camy-Peyret, C., Claveau, C., Clerbaux, C., Coheur,
P.F., Dana, V., Daumont, L., Debacker-Barilly, M.R., Flaud, J.M., Goldman, A.,
Hamdouni, A., Hess, M., Jacquemart, D., Kopke, P., Mandin, J.Y., Massie, S.,
Mikhailenko, S., Nemtchinov, V., Nikitin, A., Newnham, D., Perrin, A., Perevalov,
V.I., Regalia-Jarlot, L., Rublev, A., Schreier, F., Schult, I., Smith, K.M., Tashkun, S.A.,
Teffo, J.L., Toth, R.A., Tyuterev, V.G., Vander Auwera, J., Varanasi, P., Wagner, G.,
2005. The 2003 edition of the GEISA/IASI spectroscopic database. J. Quant.
Spectrosc. Radiat. Trans. 95, 429–467.
Karkoschka, E., Tomasko, M.G., 1992. Saturn’s upper troposphere 1986–1989. Icarus
97, 161–181.
Karkoschka, E., Tomasko, M.G., 1993. Saturn’s upper atmospheric hazes observed by
the Hubble Space Telescope. Icarus 106, 428–441.
Karkoschka, E., Tomasko, M., 2005. Saturn’s vertical and latitudinal cloud structure
1991–2004 from HST imaging in 30 filters. Icarus 179, 195–221.
Kerola, D.X., Larson, H.P., Tomasko, M.G., 1997. Analysis of the near-IR spectrum of
Saturn: A comprehensive radiative transfer model of its middle and upper
troposphere. Icarus 127, 190–212.
Kim, J.H., Kim, S.J., Geballe, T.R., Kim, S.S., Brown, L.R., 2006. High-resolution
spectroscopy of Saturn at 3 microns: CH4, CH3D, C2H2, C2H6, PH3, clouds, and
haze. Icarus 185, 476–486.
Kleiner, I., Tarrago, G., Cottaz, C., Sagui, L., Brown, L.R., Poynter, R.L., Pickett, H.M.,
Chen, P., Pearson, J.C., Sams, R.L., Blake, G.A., Matsuura, S., Nemtchinov, V.,
Varanasi, P., Fusina, L., Di Lonardo, G., 2003. NH3 and PH3 line parameters: The
2000 HITRAN update and new results. J. Quant. Spectrosc. Radiat. Trans. 82,
293–312.
Lacis, A.A., Oinas, V., 1991. A description of the correlated-k distribution method for
modelling nongray gaseous absorption, thermal emission, and multiple
scattering in vertically inhomogeneous atmospheres. J. Geophys. Res. 96 (15),
9027–9064.
Larson, H.P., 1980. Infrared spectroscopic observations of the outer planets, their
satellites, and the asteroids. Annu. Rev. Astron. Astrophys. 18, 43–75.
Lellouch, E., Bézard, B., Fouchet, T., Feuchtgruber, H., Encrenaz, T., de Graauw, T.,
2001. The deuterium abundance in Jupiter and Saturn from ISO-SWS
observations. Astron. Astrophys. 370, 610–622.
Lewis, J.S., 1969. The clouds of Jupiter and the NH3–H2O and NH3–H2S systems.
Icarus 10, 365–378.
L.N. Fletcher et al. / Icarus 214 (2011) 510–533
Lian, Y., Showman, A.P., 2008. Deep jets on gas-giant planets. Icarus 194, 597–615.
Lian, Y., Showman, A.P., 2010. Generation of equatorial jets by large-scale latent
heating on the giant planets. Icarus 207, 373–393.
Mandin, J., 1995. Line positions and intensities in the vibrational system 2m2/
m2 + m4/2m4/m1/m3 of arsine. J. Mol. Spectrosc. 172, 319–329.
Marten, A., Rouan, D., Baluteau, J.P., Gautier, D., Conrath, B.J., Hanel, R.A., Kunde, V.,
Samuelson, R., Chedin, A., Scott, N., 1981. Study of the ammonia ice cloud layer
in the equatorial region of Jupiter from the infrared interferometric experiment
on Voyager. Icarus 46, 233–248.
Martonchik, J.V., Orton, G.S., Appleby, J.F., 1984. Optical properties of NH3 ice from
the far infrared to the near ultraviolet. Appl. Opt. 23, 541–547.
McCord, T.B., Coradini, A., Hibbitts, C.A., Capaccioni, F., Hansen, G.B., Filacchione, G.,
Clark, R.N., Cerroni, P., Brown, R.H., Baines, K.H., Bellucci, G., Bibring, J.-P.,
Buratti, B.J., Bussoletti, E., Combes, M., Cruikshank, D.P., Drossart, P., Formisano,
V., Jaumann, R., Langevin, Y., Matson, D.L., Nelson, R.M., Nicholson, P.D., Sicardy,
B., Sotin, C., 2004. Cassini VIMS observations of the Galilean satellites including
the VIMS calibration procedure. Icarus. 172, 101–126.
Nixon, C.A., Irwin, P.G.J., Calcutt, S.B., Taylor, F.W., Carlson, R.W., 2001. Atmospheric
composition and cloud structure in jovian 5-lm hotspots from analysis of
Galileo NIMS measurements. Icarus 150, 48–68.
Noll, K.S., Larson, H.P., 1990. The spectrum of Saturn from 1990–2230 cm1:
Abundances of AsH3, CH3D, CO, GeH4, and PH3. Icarus 89, 168–189.
Noll, K.S., Knacke, R.F., Geballe, T.R., Tokunaga, A.T., 1986. Detection of carbon
monoxide in Saturn. Astrophys. J. 309, L91–L94.
Noll, K.S., Knacke, R.F., Geballe, T.R., Tokunaga, A.T., 1988. Evidence for germane in
Saturn. Icarus 75, 409–422.
Noll, K.S., Geballe, T.R., Knacke, R.F., 1989. Arsine in Saturn and Jupiter. Astrophys. J.
338, L71–L74.
Noll, K.S., Larson, H.P., Geballe, T.R., 1990. The abundance of AsH3 in Jupiter. Icarus
83, 494–499.
Orton, G.S., Serabyn, E., Lee, Y.T., 2000. Vertical distribution of PH3 in Saturn from
observations of its 1-0 and 3-2 rotational lines. Icarus 146, 48–59.
Orton, G.S., Baines, K.H., Cruikshank, D., Cuzzi, J.N., Krimigis, S.M., Miller, S.,
Lellouch, E., 2009. Review of knowledge prior to the Cassini-Huygens Mission
and Concurrent Research. In: Saturn from Cassini-Huygens. Springer, pp. 9–54
(Chapter 2).
Pérez-Hoyos, S., Sánchez-Lavega, A., French, R.G., Rojas, J.F., 2005. Saturn’s cloud
structure and temporal evolution from ten years of Hubble Space Telescope
images (1994–2003). Icarus 176, 155–174.
Plass, G.N., Kattawar, G.W., Catchings, F.E., 1973. Matrix operator theory of radiative
transfer. 1: Rayleigh scattering. Appl. Opt. 12, 314–329.
Porco, C.C., Baker, E., Barbara, J., Beurle, K., Brahic, A., Burns, J.A., Charnoz, S., Cooper,
N., Dawson, D.D., Del Genio, A.D., Denk, T., Dones, L., Dyudina, U., Evans, M.W.,
Giese, B., Grazier, K., Helfenstein, P., Ingersoll, A.P., Jacobson, R.A., Johnson, T.V.,
McEwen, A., Murray, C.D., Neukum, G., Owen, W.M., Perry, J., Roatsch, T., Spitale,
J., Squyres, S., Thomas, P., Tiscareno, M., Turtle, E., Vasavada, A.R., Veverka, J.,
Wagner, R., West, R., 2005. Cassini imaging science: Initial results on Saturn’s
atmosphere. Science 307, 1243–1247.
Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T., 1992. Numerical
Recipes, second ed. Cambridge University Press, Cambridge.
Read, P.L., Conrath, B.J., Fletcher, L.N., Gierasch, P.J., Simon-Miller, A.A., Zuchowski,
L.C., 2009. Mapping potential vorticity dynamics on Saturn: Zonal mean
circulation from Cassini and Voyager data. Planet. Space Sci. 57, 1682–1698.
533
Rodgers, C.D., 2000. Inverse Methods for Atmospheric Remote Sounding: Theory
and Practice. World Scientific.
Roos-Serote, M., Drossart, P., Encrenaz, T., Lellouch, E., Carlson, R.W., Baines, K.H.,
Kamp, L., Mehlman, R., Orton, G.S., Calcutt, S., Irwin, P., Taylor, F., Weir, A., 1998.
Analysis of Jupiter North Equatorial Belt hot spots in the 4–5-lm range from
Galileo/near-infrared mapping spectrometer observations: Measurements of
cloud opacity, water, and ammonia. J. Geophys. Res. 103, 23023–23042.
Rothman, L.S., Jacquemart, D., Barbe, A., Benner, D.C., Birk, M., Brown, L.R., Carleer,
M.R., Chackerian, C., Chance, K., Coudert, L.H., Dana, V., Devi, V.M., Flaud, J.M.,
Gamache, R.R., Goldman, A., Hartmann, J.M., Jucks, K.W., Maki, A.G., Mandin, J.Y.,
Massie, S.T., Orphal, J., Perrin, A., Rinsland, C.P., Smith, M.A.H., Tennyson, J.,
Tolchenov, R.N., Toth, R.A., Vander Auwera, J., Varanasi, P., Wagner, G., 2005. The
HITRAN 2004 molecular spectroscopic database. J. Quant. Spectrosc. Radiat.
Trans. 96, 139–204.
Salyk, C., Ingersoll, A., Lorre, J., Vasavada, A., Del Genio, A., 2006. Interaction between
eddies and mean flow in Jupiter’s atmosphere: Analysis of Cassini imaging data.
Icarus 185 (2), 430–442.
Showman, A.P., de Pater, I., 2005. Dynamical implications of Jupiter’s tropospheric
ammonia abundance. Icarus 174, 192–204.
Showman, A.P., Ingersoll, A.P., 1998. Interpretation of Galileo probe data and
implications for Jupiter’s dry downdrafts. Icarus 132, 205–220.
Sromovsky, L.A., Fry, P.M., 2010a. The source of 3-lm absorption in Jupiter’s clouds:
Reanalysis of ISO observations using new NH3 absorption models. Icarus 210,
211–229.
Sromovsky, L.A., Fry, P.M., 2010b. The source of widespread 3-lm absorption in
Jupiter’s clouds: Constraints from 2000 Cassini VIMS observations. Icarus 210,
230–257.
Stam, D.M., Banfield, D., Gierasch, P.J., Nicholson, P.D., Matthews, K., 2001. Near-IR
spectrophotometry of saturnian aerosols – Meridional and vertical distribution.
Icarus 152, 407–422.
Tarrago, G., Lacome, Levy, A., Guelachvili, G., Bezard, B., Drossart, P., 1992.
Phosphine spectrum at 4–5 lm: Analysis and line-by-line simulation of 2v2,
v2 + v4, v1, and v3 bands. J. Mol. Spectrosc. 154 (1), 30–42.
Temma, T., Chanover, N.J., Simon-Miller, A.A., Glenar, D.A., Hillman, J.J., Kuehn, D.M.,
2005. Vertical structure modeling of Saturn’s equatorial region using high
spectral resolution imaging. Icarus 175, 464–489.
Terrile, R.J., Westphal, J.A., 1977. The vertical cloud structure of Jupiter from
5 micron measurements. Icarus 30, 274–281.
Vasavada, A.R., Hörst, S.M., Kennedy, M.R., Ingersoll, A.P., Porco, C.C., Del Genio, A.D.,
West, R.A., 2006. Cassini imaging of Saturn: Southern hemisphere winds and
vortices. J. Geophys. Res. (Planets) 111 (E10), 5004.
Weidenschilling, S.J., Lewis, J.S., 1973. Atmospheric and cloud structures of the
jovian planets. Icarus 20, 465–476.
Weisstein, E.W., Serabyn, E., 1996. Submillimeter line search in Jupiter and Saturn.
Icarus 123, 23–36.
West, R.A., Baines, K.H., Karkoschka, E., Sánchez-Lavega, A., 2009. Clouds and
aerosols in Saturn’s atmosphere. In: Saturn from Cassini-Huygens. Springer, pp.
161–179 (Chapter 7).
Westphal, J.A., Matthews, K., Terrile, R.J., 1974. Five-micron pictures of Jupiter.
Astrophys. J. 188, L111–L112.
Yanamandra-Fisher, P.A., Orton, G.S., Fisher, B.M., Sanchez-Lavega, A., 2001. Note:
Saturn’s 5.2-lm cold spots: Unexpected cloud variability. Icarus 150, 189–193.